Patient labeling method, device, equipment and storage medium based on hierarchical calculation

Abstract

The present application discloses a patient labeling method, device, equipment and storage medium based on hierarchical calculation, which relates to the field of smart medical technology. For target patients who do not hit the high-risk diagnostic label, high-risk drug label and drug type threshold of the target disease in the target disease, the historical number of medical visits and historical medical record labels are analyzed, and hierarchical identification is used to identify in multiple dimensions to ensure comprehensive identification of risks and improve labeling accuracy. The method includes: obtaining high-risk diagnostic labels, high-risk drug labels and drug type thresholds; determining the target patient among the patients to be labeled for the target disease; counting the historical number of medical visits of the target patient in the historical time period; when the historical number of medical visits is greater than or equal to the medical number threshold, obtaining the historical medical records of the target patient, using a natural language processing algorithm to identify the historical medical records, determining the project indicators of the target patient, and marking the target patient using the risk label corresponding to the project indicator.

Description

Technical Field

[0001] The present application relates to the field of smart medical technology, and in particular to a patient labeling method, device, equipment and storage medium based on hierarchical computing. Background Art

[0002] With the continuous development of smart medical technology, people are more and more concerned about their own health problems. Many hospitals and institutions will mark patients for risk, so as to provide patients with disease treatment, product recommendations and other services. Taking diabetes as an example, the purpose of marking diabetic patients is to distinguish the risk levels of different diabetic patients, supplement diabetes risk labels, and enable the platform to analyze diabetic patients with different risk levels, so as to provide disease treatment, product recommendations and other services for patients with different risk levels in the future.

[0003] In related technologies, many smart medical platforms obtain the index thresholds for high-risk diabetic patients in medical guidelines when marking diabetic patients, and set preset conditions for screening diabetic patients based on the index thresholds. If the type of diabetes is type 1 diabetes, there is no regular medical treatment, and the blood sugar index is greater than the preset threshold, the diabetic patient will be marked as high risk. If it does not belong to the above situation, it will be marked as medium or low risk.

[0004] In the process of implementing the present invention, the inventors found that the related technology has at least the following technical problems:

[0005] When distinguishing the risk levels of diabetic patients, the platform only uses known types of diabetes and glycemic indexes for rough distinction. As a result, some diabetic patients may be in a high-risk state, but are classified as low-risk patients because they do not meet the high-risk conditions set by medicine for diabetes. This leads to incomplete risk level identification and low marking accuracy. Summary of the invention

[0006] In view of this, the present application provides a patient marking method, device, equipment and storage medium based on hierarchical calculation, the main purpose of which is to solve the current problems of incomplete risk level identification and low marking accuracy.

[0007] According to the first aspect of the present application, a patient marking method based on hierarchical calculation is provided, the method comprising:

[0008] Obtaining a high-risk diagnosis label, a high-risk drug label, and a drug type threshold, wherein the high-risk diagnosis label and the high-risk drug label are obtained in a medical guideline and are associated with the target disease, and the drug type threshold is obtained by calculating the type average of historical drug types of multiple sample patients, and the multiple sample patients are marked as high risk;

[0009] Determine a target patient among the patients to be marked with a target disease according to the high-risk diagnostic label, the high-risk drug label, and the drug type threshold, wherein the target patient is a patient among the patients to be marked who does not hit the high-risk diagnostic label, the high-risk drug label, and the drug type threshold of the target disease;

[0010] Count the number of historical medical visits of the target patient within a historical period of time;

[0011] When the historical number of medical visits is greater than or equal to the medical visit threshold, the historical medical records of the target patient are obtained, the historical medical records are identified using a natural language processing algorithm, the project indicators of the target patient are determined, and the target patient is marked using the risk label corresponding to the project indicator.

[0012] Optionally, determining a target patient among patients to be labeled with a target disease according to the high-risk diagnostic label, the high-risk drug label, and the drug type threshold comprises:

[0013] Acquire the patients to be marked of the target disease, filter the patients to be marked whose disease information is associated with the high-risk diagnostic label among the patients to be marked, and mark the filtered patients to be marked as high-risk;

[0014] Filtering the patients to be marked whose drug information is associated with the high-risk drug label among the filtered patients to be marked, and marking the filtered patients to be marked as high risk;

[0015] Counting the historical drug types of each patient to be marked among the remaining patients to be marked, filtering the patients to be marked whose historical drug types reach the drug type threshold, and marking the filtered patients to be marked as high risk;

[0016] The remaining patients to be marked after filtering are regarded as the target patients.

[0017] Optionally, the counting of historical drug types of each patient to be marked among the remaining patients to be marked includes:

[0018] For each patient to be marked among the currently remaining patients to be marked, reading a plurality of used drug information of the patient to be marked within a specified time period;

[0019] Acquire multiple preset drug types associated with the target disease, compare the multiple used drug information with the multiple preset drug types, and determine the number of preset drug types hit by the multiple used drug information;

[0020] The number of categories is used as the historical drug categories of the patient to be marked.

[0021] Optionally, after counting the number of historical medical visits of the target patient in a historical time period, the method further includes:

[0022] Determine a plurality of sample patients, query a risk label of the target disease associated with each sample patient in the plurality of sample patients, wherein the risk label is high risk or medium-low risk;

[0023] Grouping the multiple sample patients according to the risk labels, and grouping the sample patients with the same risk labels into the same group to obtain two sample patient groups;

[0024] For each of the two sample patient groups, counting the number of historical medical visits of each sample patient included in the sample patient group as the number of sample medical visits;

[0025] Based on the sample medical visits included in each sample patient group, the average medical visits of each sample patient group is calculated to obtain two average medical visits;

[0026] The two average times of medical treatment are processed by quartile method to obtain the median of the two average times of medical treatment, and the median is used as the threshold of the times of medical treatment.

[0027] Optionally, after counting the number of historical medical visits of the target patient in a historical time period, the method further includes:

[0028] When the historical number of medical visits is less than the medical visit threshold, the target patient is marked as high risk.

[0029] Optionally, the acquiring of the historical medical records of the target patient, identifying the historical medical records using a natural language processing algorithm, and determining the project indicators of the target patient include:

[0030] Obtaining the historical medical records of the target patient, dividing the historical medical records according to the time nodes indicated by the preset time intervals, and obtaining at least two medical record segments;

[0031] For each of the at least two medical record segments, use the natural language processing algorithm to perform text recognition on the medical record segment to determine a plurality of item names included in the medical record segment;

[0032] Determine a designated project name of a designated project, query the number of occurrences of the designated project name in the multiple project names, and use the number of occurrences as a segment index of the medical record segment;

[0033] Respectively counting the segment index of each of the at least two medical record segments to obtain at least two segment indexes of the at least two medical record segments;

[0034] The at least two segment indicators are used as project indicators for the target patient.

[0035] Optionally, the step of marking the target patient using the risk label corresponding to the project indicator includes:

[0036] When at least two segment indicators included in the project indicator are both greater than the indicator threshold, the target patient is marked with a risk indicator indicating a medium or low risk;

[0037] When there is a segment index lower than the index threshold value among at least two segment indexes included in the project index, the target patient is marked with a risk indicator indicating a high risk.

[0038] According to a second aspect of the present application, a patient marking device based on hierarchical calculation is provided, the device comprising:

[0039] an acquisition module, configured to acquire a high-risk diagnosis label, a high-risk drug label, and a drug type threshold, wherein the high-risk diagnosis label and the high-risk drug label are acquired in a medical guide and are associated with the target disease, and the drug type threshold is obtained by calculating the type average of historical drug types of multiple sample patients, and the multiple sample patients are marked as high risk;

[0040] A first determination module is used to determine a target patient among the patients to be marked with a target disease according to the high-risk diagnostic label, the high-risk drug label and the drug type threshold, wherein the target patient is a patient among the patients to be marked who does not hit the high-risk diagnostic label, the high-risk drug label and the drug type threshold of the target disease;

[0041] A statistical module, used to count the number of historical medical visits of the target patient within a historical time period;

[0042] The marking module is used to obtain the historical medical records of the target patient when the number of historical medical visits is greater than or equal to the medical visit number threshold, use a natural language processing algorithm to identify the historical medical records, determine the project indicators of the target patient, and mark the target patient using the risk label corresponding to the project indicator.

[0043] Optionally, the first determination module is used to obtain the patients to be marked with the target disease, filter out the patients to be marked whose disease information is associated with the high-risk diagnostic label among the patients to be marked, and mark the filtered patients to be marked as high risk; filter out the patients to be marked whose drug information is associated with the high-risk drug label among the filtered patients to be marked, and mark the filtered patients to be marked as high risk; count the historical drug types of each patient to be marked among the remaining patients to be marked, filter out the patients to be marked whose historical drug types reach the drug type threshold, and mark the filtered patients to be marked as high risk; and use the remaining patients to be marked after filtering as the target patients.

[0044] Optionally, the first determination module is used to read multiple used drug information of each patient to be marked among the currently remaining patients to be marked within a specified time period; obtain multiple preset drug types associated with the target disease, compare the multiple used drug information with the multiple preset drug types, and determine the number of preset drug types hit by the multiple used drug information; and use the number of types as the historical drug types of the patient to be marked.

[0045] Optionally, the device further comprises:

[0046] A second determination module is used to determine a plurality of sample patients, and query a risk label of the target disease associated with each sample patient in the plurality of sample patients, wherein the risk label is high risk or medium-low risk;

[0047] A grouping module, used for grouping the plurality of sample patients according to the risk labels, and classifying the sample patients with the same risk labels into the same group to obtain two sample patient groups;

[0048] The statistical module is further used for counting the number of historical medical visits of each sample patient included in each of the two sample patient groups as the number of sample medical visits;

[0049] A calculation module, used for calculating the average number of medical visits of each sample patient group based on the sample medical visits included in each sample patient group, to obtain two average number of medical visits;

[0050] The processing module is used to process the two average medical treatment times by using the quartile method to obtain the median of the two average medical treatment times, and use the median as the medical treatment times threshold.

[0051] Optionally, the marking module is further used to mark the target patient as high risk when the historical number of medical visits is less than the medical visit threshold.

[0052] Optionally, the marking module is used to obtain historical medical records of the target patient, divide the historical medical records according to time nodes indicated by preset time intervals, and obtain at least two medical record fragments; for each of the at least two medical record fragments, use the natural language processing algorithm to perform text recognition on the medical record fragment to determine multiple project names included in the medical record fragment; determine a specified project name of a specified project, query the number of occurrences of the specified project name in the multiple project names, and use the number of occurrences as a fragment index of the medical record fragment; separately count the fragment index of each of the at least two medical record fragments to obtain at least two fragment indexes of the at least two medical record fragments; and use the at least two fragment indexes as the project index of the target patient.

[0053] Optionally, the marking module is used to mark the target patient with a risk marker indicating medium or low risk when at least two fragment indicators included in the project indicators are both greater than the indicator threshold; and to mark the target patient with a risk marker indicating high risk when there is a fragment indicator lower than the indicator threshold among the at least two fragment indicators included in the project indicators.

[0054] According to a third aspect of the present application, a computer device is provided, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any one of the methods described in the first aspect when executing the computer program.

[0055] According to a fourth aspect of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of any one of the methods described in the first aspect are implemented.

[0056] Through the above technical scheme, the present application provides a patient marking method, device, equipment and storage medium based on hierarchical calculation. For target patients who do not hit the high-risk diagnostic label, high-risk drug label and drug type threshold of the target disease in the target disease, the present application counts the historical number of medical visits of the target patient in the historical time period, and when the historical number of medical visits is greater than or equal to the medical visit number threshold, obtains the historical medical records of the target patient, uses a natural language processing algorithm to identify the historical medical records, determines the project indicators of the target patient, and marks the target patient with the risk label corresponding to the project indicator, and uses hierarchical identification to identify the patient's risks in multiple dimensions such as the use of drugs, related indicators and the number of medical visits, to ensure comprehensive identification of risk levels and improve marking accuracy.

[0057] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present application. Also, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0059] Figure 1 A schematic diagram of a patient labeling method based on hierarchical calculation provided in an embodiment of the present application is shown;

[0060] Figure 2A A schematic diagram of a patient labeling method based on hierarchical calculation provided in an embodiment of the present application is shown;

[0061] Figure 2B A schematic diagram of a patient labeling method based on hierarchical calculation provided in an embodiment of the present application is shown;

[0062] Figure 3A A schematic diagram of the structure of a patient marking device based on hierarchical calculation provided in an embodiment of the present application is shown;

[0063] Figure 3B A schematic diagram of the structure of a patient marking device based on hierarchical calculation provided in an embodiment of the present application is shown;

[0064] Figure 4 A schematic diagram of the device structure of a computer device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0065] The exemplary embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present application and to fully convey the scope of the present application to those skilled in the art.

[0066] The present application embodiment provides a patient labeling method based on hierarchical calculation, such as Figure 1 As shown, the method includes:

[0067] 101. Obtain high-risk diagnostic labels, high-risk drug labels, and drug type thresholds. High-risk diagnostic labels and high-risk drug labels are obtained from medical guidelines and are associated with target diseases. Drug type thresholds are obtained by calculating the average value of historical drug types for multiple sample patients. Multiple sample patients are marked as high risk.

[0068] 102. According to high-risk diagnostic labels, high-risk drug labels and drug type thresholds, target patients are determined among the patients to be labeled with the target disease. The target patients are patients among the patients to be labeled who do not hit the high-risk diagnostic labels, high-risk drug labels and drug type thresholds of the target disease.

[0069] 103. Count the number of times the target patient has sought medical treatment during a historical period of time.

[0070] 104. When the number of historical medical visits is greater than or equal to the threshold number of medical visits, obtain the historical medical records of the target patient, use the natural language processing algorithm to identify the historical medical records, determine the project indicators of the target patient, and mark the target patient with the risk label corresponding to the project indicator.

[0071] The method provided in the embodiment of the present application counts the target patient's historical medical visits within a historical time period for a target patient who does not hit the high-risk diagnostic label, high-risk drug label, and drug type threshold of the target disease in the target disease, and when the historical number of medical visits is greater than or equal to the medical visit threshold, obtains the target patient's historical medical records, uses a natural language processing algorithm to identify the historical medical records, determines the target patient's project indicators, marks the target patient with the risk label corresponding to the project indicator, and uses a hierarchical identification method to identify the patient's risks in multiple dimensions such as the use of drugs, related indicators, and the number of medical visits, to ensure comprehensive identification of risk levels and improve marking accuracy.

[0072] The present application embodiment provides a patient labeling method based on hierarchical calculation, such as Figure 2A As shown, the method includes:

[0073] 201. Identify target patients among patients with the target disease to be labeled.

[0074] Diabetes is not as clearly distinguished as other diseases in terms of risk grading. In clinical terms, diabetes can be divided into type 1 diabetes and type 2 diabetes. Since in real life, patients with type 1 diabetes are more susceptible to the disease than patients with type 2 diabetes and often suffer from high-risk complications, patients with type 1 diabetes are at high risk for patients with type 2 diabetes. At present, some online platforms usually judge whether patients are at high risk of diabetes based on whether they are type 1 diabetes patients. If the patient is type 1 diabetes, he is marked as high risk, otherwise he is marked as medium-low risk. However, the applicant recognizes that this way of marking patients is relatively rough. It is very likely that some diabetic patients are already at high risk in terms of medication, certain indicators or medical treatment, but because they do not meet the above-mentioned judgment conditions for diabetes, they are classified as medium- and low-risk patients, resulting in incomplete risk level identification and low marking accuracy.

[0075] Therefore, the present application proposes a patient marking method, which can be applied to a smart medical platform, so that the smart medical platform can count the target patients' historical medical visits in the historical time period for the target patients who do not hit the high-risk diagnostic label, high-risk drug label and drug type threshold of the target disease in the target disease, and when the historical medical visits are greater than or equal to the medical visits threshold, obtain the target patient's historical medical records, use a natural language processing algorithm to identify the historical medical records, determine the project indicators of the target patients, and use the risk labels corresponding to the project indicators to mark the target patients, and use a hierarchical identification method to identify the patient's risk in multiple dimensions such as the use of drugs, related indicators and the number of medical visits, so as to ensure comprehensive identification of risk levels and improve the marking accuracy. Among them, the smart medical platform can be run based on an independent server, or it can be run based on a server that provides cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDNs), and basic cloud computing such as big data and artificial intelligence platforms. In this way, when the relevant information of the target patient is subsequently identified, the medical data of the target patient, such as personal health records, prescriptions, examination reports, etc., can be downloaded from the server.

[0076] Among them, the target patients are those who do not hit the high-risk diagnostic label, high-risk drug label and drug type threshold of the target disease among the patients to be marked. In other words, many high-risk patients can be identified through some existing identification rules. These identified patients who are obviously high-risk can be directly marked as high-risk, and patients who do not obviously hit the high-risk situation are used as target patients for subsequent multi-dimensional identification. In order to identify such patients who are obviously high-risk, the smart medical platform will obtain high-risk diagnostic labels and high-risk drug labels that are associated with the target disease in the medical guidelines in advance, and determine multiple sample patients marked as high-risk, calculate the average value of the historical drug types of multiple sample patients to obtain the drug type threshold, and then use the high-risk diagnostic labels, high-risk drug labels and drug type threshold to determine the target patients for subsequent multi-dimensional identification. The specific process of determining the target patients among the patients to be marked with the target disease is as follows:

[0077] First, the patients to be marked of the target disease are obtained, and according to the high-risk diagnosis label, high-risk drug label and drug type threshold, the patients to be marked whose disease information is associated with the high-risk diagnosis label are filtered out among the patients to be marked, and the filtered patients to be marked are marked as high risk. Specifically, two methods can be used to identify disease information. One method is to obtain the patient data information of the patient to be marked, identify the patient data information by text recognition, extract the disease type in the patient data information as the target disease type, and take the patients to be marked who are associated with the high-risk diagnosis label of the target disease in the target disease type as the target patient. It should be noted that since the patients to be marked may also go to the hospital for other diseases, such as colds, appendicitis, etc., this will cause the patient data information to have disease types that are not related to the target disease. Therefore, the disease types known to be related to the target disease in the medical guide can be used to filter all the disease types of the patients to be marked, and the disease types related to the target disease in the patient data information are taken as the target disease types. Another method is to identify whether the patient to be marked meets the high-risk diagnostic label of the target disease in the ICD (International Classification of Diseases) disease code of the patient to be marked, and filter the patient to be marked when it is determined to be in compliance. For example, taking diabetes as an example, identify the patients to be marked who meet the high-risk diagnostic labels such as type 1 diabetes, special types of diabetes, diabetes with complications, secondary hypertension, cerebrovascular disease, heart disease, kidney disease, peripheral vascular disease, hypertensive heart disease, hypertensive kidney disease, hypertensive heart and kidney disease, hypoglycemia, hyperlipidemia, obesity, etc. in the ICD disease code, and filter these patients to be marked. Furthermore, since patients with these diseases are already at high risk, no subsequent evaluation is required, so these filtered patients to be marked are directly marked as high risk.

[0078] Subsequently, the patients to be marked whose drug information is associated with a high-risk drug label are filtered out among the filtered patients to be marked, and the filtered patients to be marked are marked as high risk. Specifically, the drug information used to treat the target disease in the patient data information can be extracted by text recognition, including the types of drugs filled in the doctor's order corresponding to the target disease and the types of drugs purchased in the pharmacy or hospital using the medical insurance card. In addition, in the actual operation process, the extracted drug types can be judged in combination with the medical guide, and the drugs described in the medical guide that can treat the target disease are retained as drug information. In this way, the patients to be marked who are associated with a high-risk drug label in the drug information are directly filtered as high-risk patients. For example, continuing to take diabetes as an example, the use of diabetes drugs, cerebrovascular disease drugs, heart disease drugs, kidney disease drugs, peripheral vascular disease drugs, retinopathy drugs, and diabetic emergency drugs are identified among patients with diabetes, and these patients to be marked are filtered. Further, since the patients who use these drugs are already at high risk, no subsequent evaluation is required, so these filtered patients to be marked are directly marked as high risk.

[0079] Next, considering that patients who use multiple drugs related to the target disease are actually in a more serious condition, they should be marked as high risk. Therefore, it is necessary to count the historical drug types of each patient to be marked among the remaining patients to be marked, filter the patients to be marked whose historical drug types reach the drug type threshold, and mark the filtered patients to be marked as high risk. Taking any patient to be marked among the remaining patients to be marked as an example, the process of counting historical drug types is as follows: First, for each patient to be marked among the remaining patients to be marked, read multiple used drug information of the patient to be marked within a specified time period, such as counting the used drug information of the patient to be marked in the past year. Subsequently, obtain multiple preset drug types associated with the target disease, compare the multiple used drug information with the multiple preset drug types, determine the number of preset drug types hit by the multiple used drug information, and use the number of types as the historical drug types of the patient to be marked. That is, query how many drugs related to the target disease the patient to be marked has used in the past year, and use the number of drug types counted as the historical drug types. After determining the historical drug types, the patients to be marked whose historical drug types reach the drug type threshold can be filtered, and the filtered patients to be marked can be marked as high risk. Continuing with diabetes as an example, the preset drug types for diabetes medications include insulin secretagogues, α-glucosidase inhibitors, biguanides, GLP receptor agonists, etc. In this way, if the historical drug types of the patient to be marked hit the above 3 categories, and the drug type threshold is also 3, the patient to be marked will be marked as high risk. Among them, it should be noted that the drug type threshold can be set according to the indicator threshold described in the medical guidelines, or the minimum number of drug types used by diabetic patients in the historical claims information can be used as the drug type threshold.

[0080] Through the above process, the patients to be marked who are obviously at high risk are filtered and directly marked as high risk. The remaining patients to be marked after filtering are patients who need to undergo multi-dimensional risk assessment in the future. Therefore, the remaining patients to be marked after filtering are used as target patients.

[0081] 202. Count the target patient's historical medical visits in the historical time period. When the historical medical visits are greater than or equal to the medical visits threshold, execute the following step 203; when the historical medical visits are less than the medical visits threshold, execute the following step 204.

[0082] In the embodiment of the present application, considering that the risk of patients who do not seek medical treatment regularly is actually higher than that of patients who seek medical treatment regularly, the smart medical platform will count the historical number of medical visits of the target patient in the historical time period, and determine whether the target patient is at high risk according to the historical number of medical visits. Among them, when evaluating the historical number of medical visits, it is necessary to set a standard for judging whether the historical number of medical visits is sufficient, that is, the number of medical visits threshold, so that the historical number of medical visits can be compared with the number of medical visits threshold in the future, in order to determine how to mark the target patient. The specific process of generating the number of medical visits threshold is as follows:

[0083] First, multiple sample patients are determined, and the risk label of the target disease associated with each sample patient in the multiple sample patients is queried, and the risk label is high risk or medium and low risk. Multiple sample patients are grouped according to the risk label, and sample patients with the same risk label are divided into the same group to obtain two sample patient groups. Subsequently, for each sample patient group in the two sample patient groups, the historical number of medical visits of each sample patient included in the sample patient group is counted as the sample number of medical visits, and based on the sample number of medical visits included in each sample patient group, the average number of medical visits of each sample patient group is calculated to obtain two average number of medical visits. Finally, the two average number of medical visits are processed using the quartile method to obtain the median of the two average number of medical visits, and the median is used as the threshold for the number of medical visits. The above process is to group the sample patients according to high risk, medium and low risk, and then count the average number of medical visits for both the high risk group and the medium and low risk group. After obtaining the two average number of medical visits, the median of the two average number of medical visits is determined using the quartile, and the median is used as the threshold for the number of medical visits.

[0084] In this way, the historical number of medical visits of the target patient within the historical time period is counted. When the historical number of medical visits is greater than or equal to the medical visit threshold, it means that the target patient actually seeks medical treatment regularly, and the target patient cannot be directly marked as high risk, and further evaluation is required. Therefore, the following step 203 is executed; and when the historical number of medical visits is less than the medical visit threshold, it means that the target patient does not seek medical treatment regularly and there is a higher risk, and needs to be marked as high risk. Therefore, the following step 204 is executed.

[0085] 203. When the number of historical medical visits is greater than or equal to the threshold number of medical visits, the historical medical records of the target patient are obtained, the historical medical records are identified using a natural language processing algorithm, the project indicators of the target patient are determined, and the target patient is marked using the risk label corresponding to the project indicator.

[0086] In an embodiment of the present application, when the number of historical medical visits is greater than or equal to the threshold number of medical visits, it means that the target patient actually has regular medical visits, and the target patient cannot be directly marked as high-risk, and further evaluation is required. Considering that some patients only undergo some major project examinations related to the target disease in the historical process, and are not counted in the regular number of medical visits, therefore, in the continued evaluation process, it is necessary to obtain the historical medical records of the target patient, and divide the historical medical records according to the time nodes indicated by the preset time interval to obtain at least two medical record fragments. For example, the medical records of the target patient in the past two years are obtained, and the preset time interval is one year, then the medical record fragments of each year in the past two years are split. Then, for each of the at least two medical record fragments, a natural language processing algorithm is used to perform text recognition on the medical record fragment to determine the names of multiple items included in the medical record fragment. Among them, the natural language processing algorithm can be an NLP algorithm (Neuro-Linguistic Programming), which obtains all the project names as samples, trains the samples with the NLP algorithm, and obtains a language recognition model. After the medical record fragment is input into the language recognition model, multiple project names in the medical record fragment can be directly output, or the project names can be directly marked in yellow in the medical record fragment. Subsequently, the specified project name of the specified project is determined, and the number of occurrences of the specified project name in multiple project names is queried, and the number of occurrences is used as the fragment index of the medical record fragment. For example, for diabetic patients, the specified project names are "blood sugar" and "glycosylated hemoglobin (HBA1C)", and the number of occurrences of "blood sugar" and "glycosylated hemoglobin (HBA1C)" in the medical record fragment is counted as a fragment index, that is, the project name is glycosylated hemoglobin (HBA1C) or blood sugar counted once.

[0087] By repeatedly executing the above process, the fragment index of each medical record fragment in at least two medical record fragments can be counted separately to obtain at least two fragment indexes of at least two medical record fragments, and the at least two fragment indexes are used as the project index of the target patient. After determining the project index of the target patient, when at least two fragment indexes included in the project index are greater than the index threshold, it means that the target patient actually pays attention to the examination of the target disease and the risk is low. Therefore, the target patient is marked with a risk mark indicating medium and low risks. When there is a fragment index lower than the index threshold among the at least two fragment indexes included in the project index, it means that the target patient does not pay attention to daily examinations and the risk of the target patient is high. Therefore, the target patient is marked with a risk mark indicating high risk.

[0088] 204. When the number of historical medical visits is less than the medical visit threshold, the target patient is marked as high risk.

[0089] In an embodiment of the present application, when the number of historical medical visits is less than the medical visit threshold, it means that the target patient does not have regular medical visits and there is a higher risk, and needs to be marked as high risk. Therefore, the target patient is marked as high risk.

[0090] In summary, taking diabetes as an example, the technical process of this application is as follows:

[0091] See also Figure 2B Among the patients to be marked with diabetes, those who meet the high-risk diagnostic labels such as type 1 diabetes, special type diabetes, diabetes with complications, secondary hypertension, cerebrovascular disease, heart disease, kidney disease, peripheral vascular disease, hypertensive heart disease, hypertensive kidney disease, hypertensive heart and kidney disease, hypoglycemia, hyperlipidemia, and obesity are filtered. Next, the patients to be marked who use diabetes drugs, cerebrovascular disease drugs, heart disease drugs, kidney disease drugs, peripheral vascular disease drugs, retinopathy drugs, and diabetic emergency drugs are filtered. Finally, the patients to be marked whose historical drug types reach the drug type threshold are filtered, and the remaining patients to be marked are used as target patients. Next, the historical number of medical visits of the target patient in the historical time period is counted. When the historical number of medical visits is greater than or equal to the medical number threshold, the target patient is a regular medical visit. Therefore, the number of blood sugar and glycosylated hemoglobin (HBA1C) checks of the target patient each year is continued to be calculated, and when the number is greater than or equal to 1, the target patient is marked as medium-low risk, and when the number is equal to 0, the target patient is marked as high risk. When the historical number of medical visits is less than the threshold number of medical visits, the target patient has irregular medical visits and is marked as high risk.

[0092] The method provided in the embodiment of the present application counts the target patient's historical medical visits within a historical time period for a target patient who does not hit the high-risk diagnostic label, high-risk drug label, and drug type threshold of the target disease in the target disease, and when the historical number of medical visits is greater than or equal to the medical visit threshold, obtains the target patient's historical medical records, uses a natural language processing algorithm to identify the historical medical records, determines the target patient's project indicators, marks the target patient with the risk label corresponding to the project indicator, and uses a hierarchical identification method to identify the patient's risks in multiple dimensions such as the use of drugs, related indicators, and the number of medical visits, to ensure comprehensive identification of risk levels and improve marking accuracy.

[0093] Further, as Figure 1 In a specific implementation of the method, the present application provides a patient marking device based on hierarchical calculation, such as Figure 3A As shown, the device includes: an acquisition module 301, a first determination module 302, a statistics module 303, and a marking module 304.

[0094] The acquisition module 301 is used to acquire a high-risk diagnosis label, a high-risk drug label and a drug type threshold, wherein the high-risk diagnosis label and the high-risk drug label are acquired in a medical guide and are associated with the target disease, and the drug type threshold is obtained by calculating the average value of historical drug types of multiple sample patients, and the multiple sample patients are marked as high risk;

[0095] The first determination module 302 is used to determine a target patient among the patients to be marked of the target disease according to the high-risk diagnostic label, the high-risk drug label and the drug type threshold, wherein the target patient is a patient among the patients to be marked who does not hit the high-risk diagnostic label, the high-risk drug label and the drug type threshold of the target disease;

[0096] The statistical module 303 is used to count the number of historical medical visits of the target patient within a historical time period;

[0097] The marking module 304 is used to obtain the historical medical records of the target patient when the historical number of medical visits is greater than or equal to the medical visit number threshold, use a natural language processing algorithm to identify the historical medical records, determine the project indicators of the target patient, and mark the target patient using the risk label corresponding to the project indicator.

[0098] In a specific application scenario, the first determination module 302 is used to obtain the patients to be marked of the target disease, filter out the patients to be marked whose disease information is associated with the high-risk diagnostic label among the patients to be marked, and mark the filtered patients to be marked as high risk; filter out the patients to be marked whose drug information is associated with the high-risk drug label among the filtered patients to be marked, and mark the filtered patients to be marked as high risk; count the historical drug types of each patient to be marked among the remaining patients to be marked, filter out the patients to be marked whose historical drug types reach the drug type threshold, and mark the filtered patients to be marked as high risk; and use the remaining patients to be marked after filtering as the target patients.

[0099] In a specific application scenario, the first determination module 302 is used to read multiple used drug information of each patient to be marked among the currently remaining patients to be marked within a specified time period; obtain multiple preset drug types associated with the target disease, compare the multiple used drug information with the multiple preset drug types, and determine the number of preset drug types hit by the multiple used drug information; and use the number of types as the historical drug types of the patient to be marked.

[0100] In specific application scenarios, such as Figure 3BAs shown, the device further includes: a second determination module 305 , a grouping module 306 , a calculation module 307 and a processing module 308 .

[0101] The second determination module 305 is used to determine a plurality of sample patients, and query the risk label of the target disease associated with each sample patient in the plurality of sample patients, wherein the risk label is high risk or medium-low risk;

[0102] The grouping module 306 is used to group the multiple sample patients according to the risk labels, and divide the sample patients with the same risk labels into the same group to obtain two sample patient groups;

[0103] The statistical module 303 is further used to count the number of historical medical visits of each sample patient included in each of the two sample patient groups as the number of sample medical visits;

[0104] The calculation module 307 is used to calculate the average number of medical visits for each sample patient group based on the sample medical visits included in each sample patient group, and obtain two average medical visits;

[0105] The processing module 308 is used to process the two average medical visits by using the quartile method to obtain the median of the two average medical visits, and use the median as the medical visits threshold.

[0106] In a specific application scenario, the marking module 304 is also used to mark the target patient as high risk when the historical number of medical visits is less than the medical visit threshold.

[0107] In a specific application scenario, the marking module 304 is used to obtain the historical medical records of the target patient, divide the historical medical records according to the time nodes indicated by the preset time interval, and obtain at least two medical record fragments; for each of the at least two medical record fragments, use the natural language processing algorithm to perform text recognition on the medical record fragment, and determine the multiple project names included in the medical record fragment; determine the specified project name of the specified project, query the number of occurrences of the specified project name in the multiple project names, and use the number of occurrences as the fragment index of the medical record fragment; respectively count the fragment index of each of the at least two medical record fragments to obtain at least two fragment indexes of the at least two medical record fragments; use the at least two fragment indexes as the project index of the target patient.

[0108] In a specific application scenario, the marking module 304 is used to mark the target patient with a risk marker indicating medium or low risk when at least two fragment indicators included in the project indicator are greater than the indicator threshold; when there is a fragment indicator lower than the indicator threshold among the at least two fragment indicators included in the project indicator, the target patient is marked with a risk marker indicating high risk.

[0109] The device provided in the embodiment of the present application counts the target patient's historical medical visits within a historical time period for a target patient who does not hit the high-risk diagnostic label, high-risk drug label, and drug type threshold of the target disease in the target disease, and when the historical number of medical visits is greater than or equal to the medical visit threshold, obtains the target patient's historical medical records, uses a natural language processing algorithm to identify the historical medical records, determines the target patient's project indicators, marks the target patient with the risk label corresponding to the project indicator, and uses a hierarchical identification method to identify the patient's risks in multiple dimensions such as the use of drugs, related indicators, and the number of medical visits, to ensure comprehensive identification of risk levels and improve marking accuracy.

[0110] It should be noted that for other corresponding descriptions of the functional units involved in the patient marking device based on hierarchical calculation provided in the embodiment of the present application, reference can be made to Figure 1 and FIG. 2A to FIG. 2B The corresponding description in will not be repeated here.

[0111] In an exemplary embodiment, see Figure 4 , and also provides a device, which includes a bus, a processor, a memory and a communication interface, and may also include an input and output interface and a display device, wherein each functional unit can communicate with each other through the bus. The memory stores a computer program, and the processor is used to execute the program stored in the memory and execute the patient marking method based on hierarchical calculation in the above embodiment.

[0112] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the patient marking method based on hierarchical calculation.

[0113] Through the description of the above implementation methods, those skilled in the art can clearly understand that the present application can be implemented by hardware, or by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.), including a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each implementation scenario of the present application.

[0114] Those skilled in the art will appreciate that the accompanying drawings are merely schematic diagrams of a preferred implementation scenario, and the modules or processes in the accompanying drawings are not necessarily required for implementing the present application.

[0115] Those skilled in the art will appreciate that the modules in the device in the implementation scenario can be distributed in the device in the implementation scenario according to the implementation scenario description, or can be changed accordingly and located in one or more devices different from the implementation scenario. The modules in the above implementation scenario can be combined into one module, or can be further split into multiple sub-modules.

[0116] The above application serial numbers are for description only and do not represent the advantages or disadvantages of the implementation scenarios.

[0117] The above disclosure only discloses several specific implementation scenarios of the present application. However, the present application is not limited thereto, and any changes that can be conceived by technicians in this field should fall within the scope of protection of the present application.

Claims

1. A patient labeling method based on hierarchical calculation, characterized in that: include: Obtaining a high-risk diagnosis label, a high-risk drug label, and a drug type threshold, wherein the high-risk diagnosis label and the high-risk drug label are obtained in a medical guideline and are associated with a target disease, and the drug type threshold is obtained by calculating the type average of historical drug types of multiple sample patients, and the multiple sample patients are marked as high risk; Determine a target patient among the patients to be marked of the target disease according to the high-risk diagnostic label, the high-risk drug label and the drug type threshold, wherein the target patient is a patient among the patients to be marked who does not hit the high-risk diagnostic label, the high-risk drug label and the drug type threshold of the target disease, wherein when the historical drug type of the patient to be marked does not reach the drug type threshold, it is determined that the patient to be marked does not hit the drug type threshold; Count the number of historical medical visits of the target patient within a historical period of time; When the number of historical medical visits is greater than or equal to the threshold number of medical visits, the historical medical records of the target patient are obtained, the historical medical records are identified by a natural language processing algorithm, the project index of the target patient is determined, and the target patient is marked by a risk label corresponding to the project index, wherein when determining the project index of the target patient, the historical medical records of the target patient are obtained, and the historical medical records are divided according to the time nodes indicated by the preset time interval to obtain at least two medical record segments; for each of the at least two medical record segments, the natural language processing algorithm is used to perform text recognition on the medical record segment to determine the multiple project names included in the medical record segment; the designated project name of the designated project is determined, and the number of occurrences of the designated project name in the multiple project names is queried, and the number of occurrences is used as the segment index of the medical record segment; the segment index of each of the at least two medical record segments is respectively counted to obtain at least two segment indexes of the at least two medical record segments; the at least two segment indexes are used as the project index of the target patient; Among them, when the target patient is marked using the risk label corresponding to the project indicator, when at least two fragment indicators included in the project indicator are greater than the indicator threshold, the target patient is marked with a risk identifier indicating medium or low risk; when there is a fragment indicator lower than the indicator threshold among the at least two fragment indicators included in the project indicator, the target patient is marked with a risk identifier indicating high risk.

2. The method according to claim 1, characterized in that The step of determining a target patient from among patients to be labeled with a target disease according to the high-risk diagnostic label, the high-risk drug label, and the drug type threshold comprises: Acquire the patients to be marked of the target disease, filter the patients to be marked whose disease information is associated with the high-risk diagnostic label among the patients to be marked, and mark the filtered patients to be marked as high-risk; Filtering the patients to be marked whose drug information is associated with the high-risk drug label among the filtered patients to be marked, and marking the filtered patients to be marked as high risk; Counting the historical drug types of each patient to be marked among the remaining patients to be marked, filtering the patients to be marked whose historical drug types reach the drug type threshold, and marking the filtered patients to be marked as high risk; The remaining patients to be marked after filtering are regarded as the target patients.

3. The method according to claim 2, characterized in that The counting of the historical drug types of each patient to be marked among the remaining patients to be marked includes: For each patient to be marked among the currently remaining patients to be marked, reading a plurality of used drug information of the patient to be marked within a specified time period; Acquire multiple preset drug types associated with the target disease, compare the multiple used drug information with the multiple preset drug types, and determine the number of preset drug types hit by the multiple used drug information; The number of categories is used as the historical drug categories of the patient to be marked.

4. The method according to claim 1, characterized in that: After counting the number of historical medical visits of the target patient within the historical time period, the method further includes: Determine a plurality of sample patients, query a risk label of the target disease associated with each sample patient in the plurality of sample patients, wherein the risk label is high risk or medium-low risk; Grouping the multiple sample patients according to the risk labels, and grouping the sample patients with the same risk labels into the same group to obtain two sample patient groups; For each of the two sample patient groups, counting the number of historical medical visits of each sample patient included in the sample patient group as the number of sample medical visits; Based on the sample medical visits included in each sample patient group, the average medical visits of each sample patient group is calculated to obtain two average medical visits; The two average times of medical treatment are processed by quartile method to obtain the median of the two average times of medical treatment, and the median is used as the threshold of the times of medical treatment.

5. The method according to claim 1, characterized in that After counting the number of historical medical visits of the target patient within the historical time period, the method further includes: When the historical number of medical visits is less than the medical visit threshold, the target patient is marked as high risk.

6. A patient marking device based on hierarchical calculation, characterized in that: include: an acquisition module, for acquiring a high-risk diagnosis label, a high-risk drug label, and a drug type threshold, wherein the high-risk diagnosis label and the high-risk drug label are obtained in a medical guide and are associated with a target disease, and the drug type threshold is obtained by calculating the type average of historical drug types of a plurality of sample patients, and the plurality of sample patients are marked as high risk; A first determination module is used to determine a target patient among the patients to be marked of a target disease according to the high-risk diagnostic label, the high-risk drug label and the drug type threshold, wherein the target patient is a patient among the patients to be marked who does not hit the high-risk diagnostic label, the high-risk drug label and the drug type threshold of the target disease, wherein when the historical drug type of the patient to be marked does not reach the drug type threshold, it is determined that the patient to be marked does not hit the drug type threshold; A statistical module, used to count the number of historical medical visits of the target patient within a historical time period; A marking module, used for obtaining the historical medical records of the target patient when the number of historical medical visits is greater than or equal to a threshold number of medical visits, identifying the historical medical records using a natural language processing algorithm, determining the project index of the target patient, and marking the target patient using a risk label corresponding to the project index; wherein the marking module is used for obtaining the historical medical records of the target patient, dividing the historical medical records according to the time nodes indicated by a preset time interval, and obtaining at least two medical record segments; for each of the at least two medical record segments, performing text recognition on the medical record segment using the natural language processing algorithm, and determining a plurality of project names included in the medical record segment; determining a designated project name of a designated project, querying the number of occurrences of the designated project name in the plurality of project names, and using the number of occurrences as the segment index of the medical record segment; respectively counting the segment index of each of the at least two medical record segments, and obtaining at least two segment indexes of the at least two medical record segments; and using the at least two segment indexes as the project index of the target patient; Among them, the marking module is used to mark the target patient with a risk marker indicating medium or low risk when at least two fragment indicators included in the project indicator are greater than the indicator threshold; when there is a fragment indicator lower than the indicator threshold among the at least two fragment indicators included in the project indicator, the target patient is marked with a risk marker indicating high risk.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images