Blood donation return prediction method and electronic equipment

Abstract

The invention provides a blood donation return prediction method and electronic equipment, and relates to the technical field of blood donation behavior analysis. Comprising the steps that a blood donor data set is acquired, and the blood donor data set comprises at least one piece of blood donation record data of each blood donor; constructing blood donation survival analysis data of each blood donor according to the blood donor data set; according to the blood donation survival analysis data of each blood donor, generating a blood donation return estimation result corresponding to each observation time point; and determining at least one blood donation return analysis result according to the blood donation return estimation result corresponding to each observation time point. The survival analysis data of each blood donor is constructed according to the blood donation record data of each blood donor, and blood donation return behaviors of different blood donor groups at each observation time point can be analyzed based on the survival analysis data, so that blood donation groups with high potential blood donation are analyzed and obtained, and the blood donation groups are used for carrying out recall guidance in subsequent blood donation recall. Analysis is carried out based on real historical data, and the accuracy of an analysis result can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of blood donation behavior analysis, and in particular to a blood donation return prediction method and electronic equipment. Background Art

[0002] The stability and safety of the blood supply are highly dependent on the continuous contribution of regular blood donors. Analyzing the blood donation data of blood donors can identify groups of blood donors with high return potential and predict the return behavior of individual blood donors, thereby guiding blood donation management agencies to optimize blood donor recall and retention strategies to improve the stability of the blood supply.

[0003] Currently, the analysis of blood donors' return behavior is often achieved by aggregating the blood donation records of blood donors over many years into a single record, and using whether the blood donors respond to the blood donation return recruitment text message as the dependent variable, and using regression and machine learning methods to model.

[0004] However, the above methods lack analysis of the time dimension, and the analysis results for the blood donor group are less accurate. Summary of the Invention

[0005] The purpose of this application is to address the deficiencies in the above-mentioned prior art and provide a blood donation return prediction method and electronic equipment, so as to facilitate the prediction of average blood donation return of blood donor groups, the identification of blood donor groups with high blood donation potential, the prediction of individual blood donation return, etc., and assist in formulating blood donor recall and retention strategies.

[0006] To achieve the above objectives, the technical solutions adopted in the embodiments of the present application are as follows: In a first aspect, an embodiment of the present application provides a blood donation return prediction method, comprising: Acquire a blood donor data set, wherein the blood donor data set includes at least one blood donation record data of each blood donor, and the blood donation record data includes: basic information and blood donation related information; Constructing blood donation survival analysis data for each blood donor based on the blood donor data set, wherein the blood donation survival analysis data includes: a survival analysis variable and a repeated event survival analysis variable, wherein the survival analysis variable is composed of an event occurrence time and an event flag, and the repeated event survival analysis variable is composed of a start time, an end time, and an event flag; Generate blood donation return estimation results corresponding to each observation time point based on the blood donation survival analysis data of each blood donor; the blood donation return estimation results include: average number of blood donation returns; At least one blood donation return analysis result is determined based on the blood donation return estimation results corresponding to each observation time point.

[0007] In a second aspect, an embodiment of the present application further provides a blood donation return prediction device, comprising: an acquisition module, a data construction module, a generation module, and a determination module; The acquisition module is used to acquire a blood donor data set, wherein the blood donor data set includes at least one blood donation record data of each blood donor, and the blood donation record data includes: basic information and blood donation related information; The data construction module is used to construct blood donation survival analysis data of each blood donor based on the blood donor data set, wherein the blood donation survival analysis data includes: survival analysis variables and repeated event survival analysis variables, wherein the survival analysis variables are composed of event occurrence time and event flag, and the repeated event survival analysis variables are composed of start time, end time and event flag; The generating module is used to generate the blood donation return estimation results corresponding to each observation time point based on the blood donation survival analysis data of each blood donor; the blood donation return estimation results include: the average number of blood donation returns; The determination module is used to determine at least one blood donation return analysis result based on the blood donation return estimation results corresponding to each observation time point.

[0008] In a third aspect, an embodiment of the present application provides an electronic device, on which an automation service and a calling interface of the automation service are deployed; In response to the input blood donation record data of the blood donor, calling the automation service through the calling interface; The automated service is used to execute the blood donation return prediction method provided in the first aspect when running.

[0009] In a fourth aspect, an embodiment of the present application provides an electronic device comprising: a processor, a storage medium and a bus, wherein the storage medium stores machine-readable instructions executable by the processor. When the electronic device is running, the processor and the storage medium communicate through the bus, and the processor executes the machine-readable instructions to implement the blood donation return prediction method provided in the first aspect.

[0010] In a fifth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, executes the blood donation return prediction method provided in the first aspect.

[0011] The beneficial effects of this application are: The present application provides a blood donation return prediction method and electronic device, comprising: obtaining a blood donor data set, the blood donor data set including at least one blood donation record data of each blood donor; constructing blood donation survival analysis data of each blood donor based on the blood donor data set; generating blood donation return estimation results corresponding to each observation time point based on the blood donation survival analysis data of each blood donor; and determining at least one blood donation return analysis result based on the blood donation return estimation results corresponding to each observation time point. This method collects blood donation record data from each blood donor during the historical study period to form a blood donation data set, and then constructs survival analysis data for each blood donor based on the blood donation record data of each blood donor; by classifying each blood donor into blood donation groups, and based on the survival analysis data of each blood donor in each blood donation group, generates blood donation return estimation results corresponding to different observation time points. The blood donation return estimation results represent the average number of blood donation returns of blood donors at the observation time point. Based on the blood donation return estimation results, the blood donation return behavior of different blood donor groups at each observation time point can be analyzed, thereby analyzing blood donor groups with high blood donation potential for recall guidance in subsequent blood donation recalls. Among them, since each blood donor is a user who actually donated blood during the historical study period, and the blood donation record data of the blood donor is also real record data generated based on the blood donor's blood donation behavior, the blood donation return behavior of real blood donors can be analyzed based on real historical data, and effective blood donation return analysis results can be obtained. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without creative work.

[0013] Figure 1 A flow chart of a blood donation return prediction method provided in an embodiment of the present application; Figure 2 A flow chart of another blood donation return prediction method provided in an embodiment of the present application; Figure 3 A flow chart of another blood donation return prediction method provided in an embodiment of the present application; Figure 4 A flow chart of another blood donation return prediction method provided in an embodiment of the present application; Figure 5 A flow chart of another blood donation return prediction method provided in an embodiment of the present application; Figure 6A flow chart of another blood donation return prediction method provided in an embodiment of the present application; Figure 7 A flow chart of another blood donation return prediction method provided in an embodiment of the present application; Figure 8 A flow chart of another blood donation return prediction method provided in an embodiment of the present application; Figure 9 A flow chart of another blood donation return prediction method provided in an embodiment of the present application; Figure 10 A flow chart of another blood donation return prediction method provided in an embodiment of the present application; Figure 11 A schematic diagram of a blood donation return analysis curve provided in an embodiment of the present application; Figure 12 A schematic diagram of a blood donation return analysis curve for a multi-donor group provided in an embodiment of the present application; Figure 13 A flow chart of another blood donation return prediction method provided in an embodiment of the present application; Figure 14 A schematic diagram of an electronic device provided in an embodiment of the present application; Figure 15 A schematic diagram of a blood donation return prediction device provided in an embodiment of the present application; Figure 16 A schematic diagram of another electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0014] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. It should be understood that the drawings in the present application only serve the purpose of illustration and description and are not used to limit the scope of protection of the present application. In addition, it should be understood that the schematic drawings are not drawn to scale. The flowcharts used in this application illustrate the operations implemented according to some embodiments of the present application. It should be understood that the operations of the flowcharts can be implemented out of sequence, and steps without logical context can be reversed or implemented simultaneously. In addition, those skilled in the art, under the guidance of the contents of this application, can add one or more other operations to the flowchart, or remove one or more operations from the flowchart.

[0015] In addition, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application generally described and shown in the drawings here can be arranged and designed in various configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of the present application.

[0016] It should be noted that the term "comprising" will be used in the embodiments of the present application to indicate the existence of the features declared thereafter, but does not exclude the addition of other features.

[0017] The stability and safety of blood supply are highly dependent on the continuous contribution of regular blood donors. Analyzing and identifying groups of blood donors with high return potential and predicting the return behavior of individual blood donors are crucial for blood donation management agencies to optimize blood donor recall and retention strategies.

[0018] Currently, each blood donation record of each blood donor is often used as a sample data, and whether the blood donor responds to the recruitment text message is used as a label. Regression and machine learning methods are used to model the blood donation return to predict the blood donor's return.

[0019] The above binary classification definition of labels results in a lack of flexibility in the time dimension of the prediction results; and due to the complexity of the return behavior of blood donors, the binary classification model usually finds it difficult to achieve ideal prediction results.

[0020] Based on this, the blood donation return prediction method provided by this solution collects blood donation record data from each blood donor during the historical study period to form a blood donation data set. Then, based on the blood donation record data of each blood donor, survival analysis data of each blood donor is constructed. By classifying each blood donor into blood donation groups, and based on the survival analysis data of each blood donor in each blood donation group, blood donation return estimation results corresponding to different observation time points are generated. The blood donation return estimation results represent the average number of blood donation returns of blood donors at the observation time point. Based on the blood donation return estimation results, the blood donation return behavior of different blood donor groups at each observation time point can be analyzed, thereby analyzing blood donor groups with high blood donation potential and analyzing blood donation behavior in the long or short term, which can be used for recall guidance in subsequent blood donation recalls. Among them, since each blood donor is a user who actually donated blood during the historical study period, and the blood donation record data of the blood donor is also based on the real record data generated by the blood donor's blood donation behavior, the blood donation return behavior of real blood donors can be analyzed based on real historical data, and effective blood donation return analysis results can be obtained.

[0021] Figure 1A flowchart of a blood donation return prediction method provided in an embodiment of the present application; the execution subject of this method can be a computer device, such as Figure 1 As shown, the method may include: S101. Acquire a blood donor data set, where the blood donor data set includes at least one blood donation record data of each blood donor.

[0022] Blood donation record data includes: basic information and blood donation related information.

[0023] Each donor in the blood donor data set can be a user who actually donated blood during the historical research period. Each donor's donation record data will be recorded every time they donate blood. The historical research period can be a preset time period, for example: studying the blood donation behavior of all donors between January 1, 2020, and December 31, 2023.

[0024] At least one donation record for each donor can be retrieved from the blood center database. The requirement for at least one donation record is because some donors may return to donate multiple times during the study period, generating a new record each time they donate. Other donors may only donate once during the study period, generating only one record.

[0025] Among them, blood donation record data can include basic information and blood donation related information. Basic information can refer to demographic characteristics, such as the age, gender, identity and other information of the blood donor; blood donation related information refers to the various blood donation data generated during this blood donation, such as blood donation volume, blood donation reaction, hemoglobin test value, etc.

[0026] S102. Constructing blood donation survival analysis data of each blood donor based on the blood donor data set.

[0027] The blood donation survival analysis data includes: survival analysis variables and repeated event survival analysis variables. The survival analysis variables are composed of event occurrence time and event flags, and the repeated event survival analysis variables are composed of start time, end time and event flags.

[0028] Blood donation survival analysis data of each blood donor may be constructed based on at least one blood donation record data of each blood donor.

[0029] Among them, the survival analysis variables are composed of the event time and the event flag. The event time records the interval between two adjacent blood donation events, that is, the blood donation interval; the event flag is used to indicate whether the event to be studied actually occurred, such as whether the blood donation event successfully occurred or not.

[0030] Compared to survival analysis variables, repeated event survival analysis variables have a more complex structure and better reflect the characteristics of recurring events. They are primarily composed of three elements: a start time, an end time, and an event marker. The start time records the moment each recurring event begins, allowing us to track the dynamic changes in the event's initiation. The end time specifies the point in time when the corresponding event ends. Together with the start time, they fully outline the duration of each event in the temporal dimension, helping to analyze the duration characteristics of the event.

[0031] It can be seen that survival analysis data is closely related to time. It can capture the dynamic process of events over time and is very important for studying the changing trends of blood donation behavior and the temporal patterns of blood supply.

[0032] S103. Generate blood donation return estimation results corresponding to each observation time point based on the blood donation survival analysis data of each blood donor.

[0033] Blood donation return estimates include: Average number of blood donation returns.

[0034] Optionally, based on the blood donation survival analysis data of each blood donor obtained above, a blood donation return estimation calculation can be performed to obtain the blood donation return estimation results corresponding to each observation time point.

[0035] Each observation time point may correspond to a blood donation return estimation result. In this embodiment, the blood donation return estimation result mainly includes: the average number of blood donation returns, that is, the number of blood donation return behaviors at the time of the event.

[0036] The observation time point can be in days, weeks or months. Assuming the research period is 1-2556 days, then each day can be used as an observation time point, that is, each line in the research period can be used as an observation time point.

[0037] Blood donation return refers to a non-first blood donation behavior. If a user has donated blood before and returns to donate blood again, the user is considered to have donated blood again.

[0038] In some embodiments, each blood donor can be divided into groups based on their blood donation record data. The blood donors in the blood donor data set can be divided into multiple types of blood donation groups according to a single category or a combination of multiple categories, so as to perform blood donation return analysis on the blood donation groups.

[0039] Based on the divided blood donation groups, each blood donation group can be calculated independently to obtain the blood donation return estimation results corresponding to each blood donation group at each observation time point.

[0040] S104. Determine at least one blood donation return analysis result based on the blood donation return estimation result corresponding to each observation time point.

[0041] In some embodiments, estimated results may be returned for individual blood donors' corresponding blood donations at each observation time point to analyze the individual blood donors' blood donation behaviors in different time periods.

[0042] In other embodiments, the blood donation return estimation results corresponding to the blood donation groups at each observation time point can be used to determine the blood donation return behavior of each blood donation group, so as to analyze the blood donation groups with high potential for blood donation and the blood donation groups with high potential for blood donation in the short term or long term based on the group labels, so as to guide the formulation of subsequent blood donation recall strategies.

[0043] It is worth noting that the above analysis process is based on the actual blood donation record data of each blood donor who actually donated blood during the historical research period, and each observation time point is also calculated based on the actual blood donation record data. The blood donation return analysis results obtained are true and valid, and are the real results obtained after statistical analysis of historical data. Through this analysis result, it can play a certain guiding role in the subsequent blood donation return recruitment behavior.

[0044] In summary, the blood donation return prediction method provided in this embodiment includes: obtaining a blood donor data set, the blood donor data set including at least one blood donation record data of each blood donor; constructing blood donation survival analysis data of each blood donor based on the blood donor data set; generating blood donation return estimation results corresponding to each observation time point based on the blood donation survival analysis data of each blood donor; and determining at least one blood donation return analysis result based on the blood donation return estimation results corresponding to each observation time point. This method collects blood donation record data from each blood donor during the historical study period to form a blood donation data set, and then constructs survival analysis data for each blood donor based on the blood donation record data of each blood donor; by classifying each blood donor into blood donation groups, and based on the survival analysis data of each blood donor in each blood donation group, generates blood donation return estimation results corresponding to different observation time points. The blood donation return estimation results represent the average number of blood donation returns of blood donors at the observation time point. Based on the blood donation return estimation results, the blood donation return behavior of different blood donor groups at each observation time point can be analyzed, thereby analyzing blood donor groups with high blood donation potential for recall guidance in subsequent blood donation recalls. Among them, since each blood donor is a user who actually donated blood during the historical study period, and the blood donation record data of the blood donor is also real record data generated based on the blood donor's blood donation behavior, the blood donation return behavior of real blood donors can be analyzed based on real historical data, and effective blood donation return analysis results can be obtained.

[0045] Figure 2A flowchart of another blood donation return prediction method provided in an embodiment of the present application; optionally, in step S102, constructing blood donation survival analysis data for each blood donor based on the blood donor data set includes: S201. Sort each piece of blood donation record data of each blood donor according to the blood donation time in each piece of blood donation record data of each blood donor, and generate a sorting result of the blood donation record of each blood donor.

[0046] The blood donor data set contains multiple blood donation record data. Each blood donation record data corresponds to a blood donor ID. Different blood donation record data may correspond to the same blood donor ID, that is, the same blood donor has donated blood multiple times.

[0047] All blood donation record data of the same blood donor can be divided into a group according to the blood donor identifier in each blood donation record data. Then, the blood donation record data in a group are sorted according to the blood donation time in each blood donation record data to obtain the blood donation record sorting result of a blood donor.

[0048] The blood donation record data of each blood donor are processed in this way, and the blood donation record ranking results of each blood donor can be obtained.

[0049] S202: Determine the event occurrence time and event flag corresponding to each piece of blood donation record data of the blood donor according to the sorting result of the blood donation record of the blood donor.

[0050] The event occurrence time is used to indicate the interval between two adjacent blood donations; the event flag is used to indicate whether a blood donation return behavior occurs.

[0051] In some embodiments, the event occurrence time and event flag corresponding to each piece of blood donation record data can be calculated in sequence according to the blood donation record sorting result.

[0052] The event occurrence time can be calculated based on the blood donation time of the next blood donation record data and the blood donation time of the current blood donation record data.

[0053] The event flag can be set as follows: when a blood donation return event occurs, the event flag is 1; when a blood donation return event does not occur, the event flag is 0.

[0054] S203: Constructing a survival analysis variable corresponding to each piece of blood donation record data of the blood donor according to the event occurrence time and event flag corresponding to each piece of blood donation record data.

[0055] The survival analysis variable corresponding to each blood donation record can be represented as: event occurrence time-event flag.

[0056] S204: constructing a repeated event survival analysis variable corresponding to each piece of blood donation record data of the blood donor according to the blood donation record sorting result and the event occurrence time and event flag corresponding to each piece of blood donation record data of the blood donor.

[0057] Based on the constructed survival analysis variables and combined with the sorting results of the blood donation records of the blood donors, repeated event survival analysis variables corresponding to each blood donation record data of the blood donors can be further constructed.

[0058] The repeated event survival analysis variables corresponding to the blood donation record data can be represented as: start time-end time-event flag.

[0059] Figure 3 A flowchart of another blood donation return prediction method provided in an embodiment of the present application; optionally, in step S202, determining the event occurrence time and event flag corresponding to each blood donation record data of the blood donor based on the blood donation record sorting result of the blood donor may include: S301. According to the sorting result of the blood donation records of the blood donors, it is determined whether the next blood donation record data exists in the current blood donation record data.

[0060] The blood donation record sorting results are arranged in order. The blood donation record at the end of the sorting does not have a next blood donation record because it is already the last blood donation record. All other blood donation records have a corresponding next blood donation record, and the next record here refers to the one immediately adjacent to the current one.

[0061] S302: If yes, determine the time of occurrence of the event corresponding to the current blood donation record data according to the blood donation time in the current blood donation record data and the blood donation time in the next blood donation record data, and determine the event flag corresponding to the current blood donation record data as the first flag.

[0062] The first sign indicates that blood donation return behavior has occurred.

[0063] If the next blood donation record exists for the current blood donation record, the donation time of the next blood donation record is subtracted from the donation time of the current blood donation record to obtain the event occurrence time corresponding to the current blood donation record, and the event identifier is set to the first flag. The first flag indicates that a blood donation return event has occurred, that is, the blood donor has made a repeated blood donation. The first flag can be represented by 1.

[0064] The current piece of blood donation record data may refer to any piece of blood donation record data in the blood donation record sorting result.

[0065] S303. If not, determine the event occurrence time corresponding to the current blood donation record data according to the blood donation return analysis end time and the blood donation time in the current blood donation record data, and determine the event flag corresponding to the current blood donation record data as the second flag.

[0066] The second flag indicates that a censoring event has occurred.

[0067] If the next blood donation record does not exist for the current blood donation record data, that is, the current blood donation record data is the last blood donation record data in the sorting result, the blood donation return analysis end time can be subtracted from the blood donation time of the current blood donation record data to obtain the event occurrence time corresponding to the current blood donation record data.

[0068] The end time of blood donation return analysis also refers to the end time of the research period, such as December 31, 2023 mentioned above.

[0069] At the same time, the event flag is determined to be the second flag, that is, a censored event occurs, and the second flag can be 0. A censored event refers to the inability to fully observe the survival time (or event occurrence time) of some individuals due to limitations in research design or data collection.

[0070] There are usually three possible situations for missing events. One is that the blood donor did not return to donate blood at the end of the study or dropped out of the study midway; another is that the blood donor had returned to donate blood before the start of the study, but the specific time of occurrence is unknown; and the third is that the return of blood donation occurred within a certain time period, but the specific time point is unknown.

[0071] Figure 4 A flowchart of another blood donation return prediction method provided in an embodiment of the present application; optionally, in step S204, constructing a repeated event survival analysis variable corresponding to each blood donation record of the blood donor based on the blood donation record sorting result and the event occurrence time and event flag corresponding to each blood donation record data of the blood donor may include: S401. Determine the start time and end time corresponding to each piece of blood donation record data according to the blood donation record sorting result and the event occurrence time corresponding to each piece of blood donation record data of the blood donor.

[0072] The start time is used to indicate the time when the blood donor may donate blood next time based on the time of the first blood donation; the end time is used to indicate the time when the blood donor returns to donate blood or the time when the blood donation is returned for analysis.

[0073] It is worth noting that we use relative time rather than absolute time to express the start time and end time, and use days as the time unit.

[0074] The start time and end time determine the time frame of the event and the measurement of risk exposure.

[0075] The start time of each blood donation record must be equal to the end time of the previous blood donation record to ensure that there is no overlap or gap in the time axis.

[0076] S402: Constructing a repeated event survival analysis variable corresponding to each piece of blood donation record data of the blood donor according to the start time and end time corresponding to each piece of blood donation record data and the event flag corresponding to each piece of blood donation record data.

[0077] The repeated event survival analysis variable corresponding to each blood donation record can be represented as: start time - end time - event flag. The event flag here is the same as the event flag result in the survival analysis variable constructed above.

[0078] Figure 5 A flowchart of another blood donation return prediction method provided in an embodiment of the present application; optionally, in step S401, determining the start time and end time corresponding to each blood donation record data based on the blood donation record sorting result and the event occurrence time corresponding to each blood donation record data of the blood donor may include: S501. For current blood donation record data, calculate the cumulative time of event occurrence times corresponding to the current blood donation record data and all blood donation record data before the current blood donation record data.

[0079] For each blood donation record, the cumulative time can be calculated by calculating the event occurrence time of the current blood donation record and all blood donation records before the current blood donation record. In other words, the sum of the event occurrence time from the first blood donation record to the current blood donation record is calculated.

[0080] S502. Use the accumulated time as the end time corresponding to the current blood donation record data.

[0081] The accumulated time obtained is used as the end time corresponding to the current blood donation record data.

[0082] S503: Using the end time of the previous blood donation record data corresponding to the current blood donation record data as the start time corresponding to the current blood donation record data.

[0083] The start time corresponding to the current blood donation record data is the end time corresponding to the previous blood donation record data, and the end time of the previous blood donation record data is the same as the start time of the next blood donation record data.

[0084] It is worth noting that if the current blood donation record data is the first-ranked blood donation record data in the blood donor's blood donation record sorting result, the start time corresponding to the current blood donation record data is determined to be 0.

[0085] Typically, the start time of the first blood donation record data can be set to 0 as the benchmark time, and the start time and end time of each subsequent blood donation record data are obtained based on 0. Of course, the benchmark can also be set as the start date of the study.

[0086] In addition, if the blood donor does not return blood during the follow-up period, the end time of the last blood donation record data can be set as the end time of the study.

[0087] For example, suppose a blood donor has three blood donation records. Arranged in order, the first blood donation record has an event time of day a, the second blood donation record has an event time of day b, and the third blood donation record has an event time of day c. Then, the first blood donation record has a start time of 0 and an end time of 0+a; the second blood donation record has a start time of 0+a and an end time of 0+a+b; the third blood donation record has a start time of 0+a+b and an end time of 0+a+b+c.

[0088] Survival analysis usually focuses on time intervals (end time - start time) rather than absolute time points.

[0089] Figure 6 A flowchart of another blood donation return prediction method provided in an embodiment of the present application; optionally, in step S103, generating a blood donation return estimation result corresponding to each observation time point based on the blood donation survival analysis data of each blood donor may include: S601. Classify each blood donor into a group according to at least one group classification method, and determine at least one blood donor group.

[0090] In some embodiments, different donor groups can be divided based on each donor's blood donation record data, such as gender, identity, and blood donation volume. Blood donor groups can be divided into a single category or a combination of multiple categories. For example, a single category can be divided only by gender or only by age; a combination of multiple categories can be divided by gender + age, weight + blood donation volume, etc.

[0091] S602. Based on the blood donation survival analysis data of each blood donor in each blood donor group, the blood donation return estimation is performed on each blood donor group to generate the blood donation return estimation results corresponding to each blood donor group at each observation time point.

[0092] Based on the division results, multiple blood donor groups can be obtained. Each blood donor group may contain multiple blood donation record data of multiple blood donors. The blood donation return estimation of each blood donor group can be performed based on the blood donation survival analysis data corresponding to the blood donation record data of each blood donor in the blood donor group to obtain the blood donation return estimation results corresponding to each blood donor group at each observation time point.

[0093] Figure 7 A flowchart of another blood donation return prediction method provided in an embodiment of the present application; optionally, in step S602, based on the blood donation survival analysis data of each blood donor in each blood donor group, blood donation return estimation is performed for each blood donor group, and blood donation return estimation results corresponding to each blood donor group at each observation time point are generated, which may include: S701. Sort each piece of blood donation record data of each blood donor in the blood donor group according to the end time corresponding to each piece of blood donation record data of each blood donor in the blood donor group to obtain a sorting result of the blood donation records of the blood donor group.

[0094] The blood donation record data can be sorted from small to large according to the end time. The sorting here only focuses on the end time and does not distinguish between blood donors. Adjacent sorts may be blood donation record data of different blood donors.

[0095] In some embodiments, if the event occurrence time corresponding to the first blood donation record data is the same as the event occurrence time corresponding to the second blood donation record data, the second blood donation record data is arranged behind the first blood donation record data; the event mark of the first blood donation record data is the first mark, and the event mark of the second blood donation record data is the second mark.

[0096] That is, if the event occurrence time corresponding to the blood donation record data with event flag 1 is the same as the event occurrence time corresponding to the blood donation record data with event flag 0, the blood donation record data with event flag 0 will be sorted after the blood donation record data with event flag 1. In other words, the censored event is sorted after the blood donation return event.

[0097] S702: Determine the number of designated blood donors corresponding to each observation time point according to the sorting results of the blood donation records of the blood donor group.

[0098] Designated blood donors were those who had not returned to donate blood and were under observation until each observation time point; each observation time point corresponded to one day within the study period.

[0099] According to the blood donation record sorting results of the blood donor group, the observation time point can be calculated. , the number of blood donors who have not returned to donate blood and are under observation Assuming the study period lasts 2556 days, then i can be 0-2555, with each day corresponding to a .

[0100] S703. Determine the blood donation return estimate corresponding to each observation time point according to the number of designated blood donors corresponding to each observation time point.

[0101] The blood donation return estimate is used to represent the average number of blood donation return events.

[0102] Based on the number of designated blood donors calculated at each observation time point , the estimated return value of blood donation corresponding to each observation time point can be calculated.

[0103] S704. Determine the upper and lower limits of the estimated blood donation return values ​​corresponding to each observation time point based on the number of designated blood donors corresponding to each observation time point, the event flags corresponding to each blood donation record data at each observation time point, and the estimated blood donation return values ​​at each observation time point.

[0104] In some embodiments, upper and lower bounds can also be calculated for the estimated return on donation value at each measured time point. Due to the limited data collected from donors' donation records, the calculated return on donation value estimate may experience random fluctuations. The upper and lower bounds provide a possible range for the estimated value, reflecting this uncertainty. Confidence intervals allow decision makers to assess risk and make more robust decisions.

[0105] Based on the sorting results of the blood donation records of the blood donor group, each piece of blood donation record data contained at each observation time point can be counted separately, that is, each piece of blood donation record data generated up to each observation time point.

[0106] By combining the event flags corresponding to each blood donation record data, the estimated blood donation return value at the observation time point, and the number of designated blood donors at the observation time point, the upper and lower limits of the estimated blood donation return value corresponding to the observation time point can be calculated.

[0107] S705. The estimated blood donation return value corresponding to each observation time point and the upper limit and lower limit of the estimated blood donation return value corresponding to each observation time point are used as the blood donation return estimation result corresponding to each observation time point of the blood donor group.

[0108] Then, the blood donation return estimation result corresponding to the above-mentioned observation time point may include three types of data: the upper limit and lower limit of the blood donation return estimation value corresponding to the observation time point, and the blood donation return estimation value corresponding to the observation time point.

[0109] Figure 8A flowchart of another blood donation return prediction method provided in an embodiment of the present application; optionally, in step S702, determining the number of designated blood donors corresponding to each observation time point based on the blood donation record sorting results of the blood donor group may include: S801. According to the sorting result of the blood donation records of the blood donor group, determine each piece of blood donation record data included up to the current observation time point.

[0110] Since the blood donation record sorting results are arranged based on the end time of the blood donation record data, and the end time is a relative time calculated relative to the set base time 0, the data with the same dimension as the observation point time is a data that starts from 0 and increases continuously. Therefore, based on the sorting results and the end time of each blood donation record data in the sorting results, it is possible to directly obtain the statistics of each blood donation record data included up to the current observation point time.

[0111] S802. According to the event flag corresponding to each blood donation record data and the designated blood donor calculation strategy corresponding to the event flag, the number of designated blood donors under each blood donation record data is calculated in turn, and the number of designated blood donors under the last blood donation record data is used as the number of designated blood donors corresponding to the current observation time point.

[0112] Optionally, the following formula can be used to calculate the observation time point: The number of blood donors who have not returned to donate blood and are under observation, that is, the number of designated blood donors : like Corresponding to repeated blood donation events, = ; like corresponds to a censored event, then = .

[0113] For repeated blood donation events: If the blood donor is at the current observation time point This corresponds to a repeated blood donation event, that is, the blood donor has donated blood before and donated blood again at this time, so the number of blood donors Remains unchanged and is equal to the number of blood donors at the previous observation time point .

[0114] For censored events: If the blood donor is at the current observation time point The corresponding event is the censored event, that is, the blood donor did not return to donate blood at this time point and is in an observation state (perhaps because of some reason he is unable to continue to donate blood, such as health problems, loss of contact, etc.), then the number of blood donors is Decrease by 1, which is equal to the number of blood donors at the previous observation time point -1.

[0115] Among them, blood donors at the current observation time point Whether it is a repeated blood donation event or a deleted event can be determined based on the event flag corresponding to the blood donation record data. If the event flag is the first flag, it is a repeated blood donation event; if the event flag is the second flag, it is a deleted event.

[0116] Optionally, in step S802, the number of designated blood donors under each blood donation record data is calculated in sequence according to the event flag corresponding to each blood donation record data and the designated blood donor calculation strategy corresponding to the event flag, which may include: according to the event flag corresponding to the current blood donation record data and the number of designated blood donors under the previous blood donation record data corresponding to the current blood donation record data, using the calculation strategy corresponding to the event flag to determine the number of designated blood donors under the current blood donation record data, until the number of designated blood donors under the last blood donation record data is calculated.

[0117] Due to the observation time The number of designated blood donors corresponding to each blood donation record data can be calculated in sequence, and the number of designated blood donors corresponding to each blood donation record data is calculated based on the number of designated blood donors corresponding to the previous blood donation record data until the observation time point is The number of designated blood donors corresponding to the last blood donation record data is used as the observation time point The corresponding number of designated blood donors.

[0118] Optionally, in step S703, determining the estimated blood donation return value corresponding to each observation time point based on the number of designated blood donors corresponding to each observation time point may include: determining the estimated blood donation return value corresponding to the current observation time point based on the number of designated blood donors corresponding to the current observation time point and the estimated blood donation return value of the previous observation time point corresponding to the current observation time point.

[0119] In some embodiments, the observation time point can be calculated using the following formula: Corresponding blood donation return estimate :

[0120] in, It can refer to the time point of observation; is the number of designated blood donors corresponding to the observation time point; Refers to the observation time point The corresponding blood donation return estimate; Observation time point The corresponding blood donation return estimate; for The previous time point.

[0121] That is to say, for the current observation time point, the number of designated blood donors corresponding to the current observation time point can be calculated based on the number of designated blood donors corresponding to the current observation time point and the blood donation return estimate at the previous observation time point corresponding to the current observation time point.

[0122] The current observation time point may be any observation time point except the initial observation time point. That is, the blood donation return estimate corresponding to each current observation time point may be calculated based on the blood donation return estimate corresponding to the previous observation time point.

[0123] Figure 9 A flowchart of another blood donation return prediction method provided in an embodiment of the present application; optionally, in step S704, determining the upper and lower limits of the estimated blood donation return value corresponding to each observation time point based on the number of designated blood donors corresponding to each observation time point, the event flag corresponding to each blood donation record data at each observation time point, and the estimated blood donation return value at each observation time point may include: S901. Determine the data identifier of the blood donor corresponding to each piece of blood donation record data according to the event flag corresponding to each piece of blood donation record data at the current observation time point.

[0124] The data identifier is used to indicate whether a blood donation return event occurs for the blood donor at the current observation time point.

[0125] In some embodiments, the observation time point can be calculated using the following formula: The corresponding blood donation return estimated variance :

[0126] in, Indicates the observation time point The data identifier of the blood donor corresponding to the blood donation record data in the jth article below.

[0127] The data identifier of the blood donor corresponding to the j-th blood donation record data can be determined based on the event flag corresponding to the j-th blood donation record data. When the event flag is the first flag, that is, the event flag indicates that a repeated blood donation event has occurred, the data identifier can be determined to be the first identifier, and the first identifier can be 1; when the event flag is the second flag, that is, the event flag indicates that a deletion event has occurred, the data identifier can be determined to be the second identifier, and the second identifier can be 0.

[0128] S902. Determine the estimated variance of the blood donation return corresponding to the current observation time point based on the estimated variance of the blood donation return at the previous observation time point corresponding to the current observation time point, the number of designated blood donors corresponding to the current observation time point, and the data identifiers of the blood donors corresponding to each blood donation record data at the current observation time point.

[0129] Similarly, the estimated variance of the blood donation at the last observation time point can be returned by iterative calculation , the current observation time point The number of designated blood donors , the current observation time point Data identifier of the corresponding blood donor Substitute into the calculation formula of the above blood donation return estimation variance to calculate the blood donation return estimation variance corresponding to the current observation time point .

[0130] S903. Determine the upper limit and lower limit of the estimated blood donation return value corresponding to the current observation time point according to the estimated variance of the blood donation return corresponding to the current observation time point and the estimated blood donation return value at the current observation time point.

[0131] Next, the upper and lower limits of the estimated blood donation return corresponding to the observation time point are calculated using the following formulas:

[0132]

[0133] in, Indicates the observation time point the corresponding upper limit of the estimated return value of blood donation; Indicates the observation time point the lower bound of the corresponding blood donation return estimate; is the confidence level, is the standard normal quantile.

[0134] The observation time point can be Corresponding blood donation return estimate and the estimated variance of blood donation returns Substitute into the formula and calculate the observation time points respectively The upper limit of the corresponding blood donation return estimate and the lower bound of the estimated return on blood donations .

[0135] Figure 10A flowchart of another blood donation return prediction method provided in an embodiment of the present application; optionally, in step S104, determining at least one blood donation return analysis result based on the blood donation return estimation results corresponding to each observation time point may include: S1001. Draw a blood donation return analysis curve corresponding to each blood donor group based on the blood donation return estimate corresponding to each observation time point and the upper limit and lower limit of the blood donation return estimate corresponding to each observation time point.

[0136] The blood donation return analysis curve is used to show the mapping relationship between the observation time points and the estimated blood donation return values.

[0137] In some embodiments, based on the obtained blood donation return estimation values ​​corresponding to each blood donor group at each observation time point combined with the upper and lower limits of the blood donation return estimation values, blood donation return analysis curves of different blood donor groups can be drawn respectively.

[0138] Figure 11 A blood donation return analysis curve diagram provided in an embodiment of the present application; Figure 11 As shown, the horizontal axis of the curve can represent the observation time point, and the vertical axis can represent the estimated value of blood donation return. Figure 11 What is shown is the blood donation return analysis curve of a certain blood donor group, among which the vertical coordinate of the middle solid curve represents the estimated blood donation return value; the vertical coordinate of the dotted curve above the solid curve represents the upper limit of the estimated blood donation return value; the vertical coordinate of the dotted curve below the solid curve represents the lower limit of the estimated blood donation return value.

[0139] S1002. Based on the blood donation return analysis curve corresponding to each blood donor group, determine the change information of the average blood donation return times of each blood donor group over time and the blood donor group with high blood donation return potential within a preset time period.

[0140] By analyzing the trend of the blood donation return analysis curve corresponding to the blood donor group, we can describe the changes in the average number of return blood donations of the blood donor group over time.

[0141] In some embodiments, a two-sample pseudo-score test can be used to perform hypothesis testing on the significance of differences in the estimated return results of blood donations of different blood donor groups. By comparing the estimated return results of blood donations of different blood donor groups with the hypothesis verification results, blood donor groups with high blood donation return potential in short-term and long-term time periods can be summarized.

[0142] In addition, the data of different blood donor groups can be plotted together in a curve to more clearly analyze the groups with high potential for blood donation.

[0143] Figure 12An embodiment of the present application provides a schematic diagram of a blood donation return analysis curve for a multi-donor group; assuming that the groups are divided by identity and blood donation reaction, identity A-No (i.e., the red curve) represents a person with identity A who did not have a blood donation reaction; identity A-Yes (i.e., the blue curve) represents a person with identity A who had a blood donation reaction; identity B-No represents a person with identity B who did not have a blood donation reaction; identity B-Yes represents a person with identity B who had a blood donation reaction.

[0144] Overall, the group that did not experience a blood donation reaction was more likely to return to donate blood than the group that did (higher vertical axis values ​​on the curve = more average number of return donations).

[0145] In the short term (180-365 days after blood donation), among the group that did not experience a blood donation reaction, the probability of returning to donate blood is: Identity B > Other identities > Identity A, and the peak of returning to donate blood (the curve is more inclined) for people with Identity B occurs after an interval of 180 days, and for people with Identity A it occurs when it is close to 1 year, and the other identities are relatively smooth.

[0146] Over the long term, individuals with Identity A who did not experience a blood donation reaction and those with other identities were more likely to return to donate (Estimated Return Blood Donation_2556 > 1), exhibiting an annual cyclicity (stronger for Identity A), with the curve flattening mid-year and then sloping towards the end of the year (when the interval approaches a full year). Individuals with Identity B had fewer returns to donate over the long term. This may be due to the regional nature of the study, and due to their unique identity, some individuals with Identity B tended to leave the study area after a certain period of time and no longer return to donate. For individuals with blood donation reactions, regardless of their identity, the average number of returns to donate was low (Estimated Return Blood Donation_2556 < 0.5), and no annual cyclicity was observed.

[0147] Figure 13 A flow chart of another blood donation return prediction method provided in an embodiment of the present application; optionally, the method of the present application may further include: S1301. Construct training sample data based on each blood donation record data of each blood donor in the blood donor data set and the survival analysis variables corresponding to each blood donation record data.

[0148] The above embodiment is an analysis of a blood donation group. In this embodiment, the return probability of blood donation of individual blood donors can also be predicted by training a prediction model.

[0149] Training sample data can be constructed based on each blood donation record data of each blood donor in the blood donor data set and the survival analysis variables corresponding to each blood donation record data, wherein the survival analysis variable corresponding to each blood donation record data can be used as label data to guide the fitting of the model.

[0150] S1302: Using training sample data, perform model fitting to obtain an individual blood donation return prediction model.

[0151] The individual blood donation return prediction model is used to determine the blood donation return probability and the cumulative blood donation return probability of the blood donor based on the blood donation record data of the blood donor.

[0152] In some embodiments, 80% of the training sample data may be divided into a training set and 20% may be divided into a test set; and a three-fold cross-validation method may be applied to the training set for resampling to form a stable performance evaluation.

[0153] Next, a prediction model for individual blood donation returns was fitted based on the training set. For the base learner, data was resampled using random Boosting weights, nodes were split repeatedly, and the optimal split direction was calculated using the Newton-Raphson method. The log-rank test was used as the split decision metric until no nodes needed to be split. Finally, a large number of skewed survival trees were integrated using the bagging algorithm to form a skewed decision random survival forest.

[0154] Data resampling with random bootstrap weights generates multiple bootstrap samples from the original data using random sampling with replacement. The weight of each sample is used for subsequent training. The weight of each sample can be uniformly distributed or generated according to a specific strategy (such as weighted sampling).

[0155] Node splitting, initialization: starting from the root node, all samples enter the current node; calculate the optimal split direction: use linear combination (oblique splitting) instead of axis parallel splitting. Assume that the splitting plane is w T x + b = 0, where w is the weight vector, x is the covariate, and b is the bias. Use the Newton-Raphson method to optimize the objective function and calculate the optimal w and b.

[0156] Split decision metric: Use the log-rank statistic as the splitting criterion to maximize the survival difference between different child nodes.

[0157] Conditions for stopping node splitting: The number of samples in the node is less than a preset threshold. The log-rank statistic is less than a certain threshold. The maximum tree depth is reached.

[0158] Return the final tree: When all nodes cannot be split further, return the final oblique survival tree.

[0159] To ensemble leaning survival trees using bagging, repeat the following steps N times (N is the number of trees in the forest): Step 1: Generate a bootstrap sample from the original data. Step 2: Train a leaning survival tree on this sample. Step 3: Save the trained tree.

[0160] Next, the Bayesian optimization method is used to optimize the hyperparameters of the fitted prediction model, including the number of trees, the number of node splits, the ratio of variables used in nodes, the ratio of samples used in nodes, and the minimum number of samples in nodes.

[0161] Bayesian optimization settings: Maximizing Uno's C-index (consistency index) was used as the optimization metric. Kriging (regression model) was selected as the surrogate model, and the covariance function was specified as the Matern 5 / 2 kernel function. The BFGS (Broyden-Fletcher-Goldfarb-Shanno, quasi-Newton method) algorithm was used to optimize the Kriging regression model. The Expected Improvement (EI) criterion was used as the acquisition function, and the DIRECT (DIviding RECTangles) global optimization method algorithm from the NLOpt (an open-source software library for nonlinear optimization) library was selected as the acquisition function optimizer.

[0162] Afterwards, various survival analysis average indicators can be used to evaluate the model fit. The importance of each type of data in the blood donation record data can be evaluated using analysis of variance, negation method and random permutation method.

[0163] Among them, the analysis of variance method can also be used to measure feature importance. Its specific implementation is as follows: the analysis of variance (ANOVA) table is calculated in non-leaf nodes to obtain the p-value of each predictor variable that contributes to the node (that is, the basic information in the blood donation record data and blood donation-related information). Then, ANOVA VI1 is defined as the number of times the p-value associated with the predictor variable is ≤0.01 when constructing the forest; the inverted feature importance is evaluated by inverting the coefficient of a specific predictor variable and calculating the reduction in out-of-bag (OOB) accuracy during the Boostrap process, thereby evaluating the importance of a single predictor variable; the random permutation feature importance is evaluated by randomly permuting the value of the predictor variable and calculating the reduction in OOB accuracy.

[0164] In some embodiments, the trained individual blood donation return prediction model can be used to predict the return probability of a tasked blood donor. The prediction can be performed using the donor's blood donation record data as input to the model, thereby predicting the return probability of the donor at different observation time points, and also predicting the cumulative return probability of the donor over a period of time.

[0165] The basic information of the blood donor shall include at least: the gender, weight, age, identity and education level of the blood donor; the blood donation-related information of the blood donor shall include at least: time of blood donation, hemoglobin test value, blood donation reaction, blood donation amount and place of blood donation.

[0166] This embodiment describes the specific data contained in the blood donation record data. The basic information of the blood donor can be understood as demographic information, including but not limited to: donor gender, weight, age, identity, and education level. The blood donation-related information includes but is not limited to the time of donation, hemoglobin test value, blood donation reaction, blood donation amount, and donation location.

[0167] Optionally, in step S102, before constructing the blood donation survival analysis data of each blood donor based on the blood donor data set, the following steps may be included: If there is an abnormality in the target blood donation record data of the blood donor, the blood donor age in the target blood donation record shall be corrected according to the blood donor age in other blood donation record data of the blood donor; or the blood donor age in the target blood donation record shall be corrected according to the age average of the category to which the blood donor belongs.

[0168] Before using the above-mentioned blood donation record data, errors during data entry can be adjusted and corrected in some ways.

[0169] For the age of blood donors, an age range that complies with regulations is usually set, such as 18-59 years old. If the age of the blood donor recorded in a certain blood donation record data of the blood donor is not within the above age range, it is first possible to determine whether there are other blood donation record data for the blood donor. If so, the abnormal record can be corrected based on the age of the blood donor recorded in other blood donation record data. Of course, the premise here is that the age recorded in other blood donation record data is accurate. Under the premise of accuracy, the age information with differences can be adjusted based on the interval time between the two blood donation records and the correct age information recorded. If there is no other blood donation record data for the blood donor, it can be corrected by the mean or median of the category to which the blood donor belongs. Usually, the gender of the blood donor can be used to calculate the average age or median age of all blood donors under the same gender as the age of the blood donor.

[0170] If there is an abnormality in the blood donor's weight in the blood donation record data, the abnormal blood donor's weight will be corrected according to the number of digits and the size of the first digit of the abnormal blood donor's weight.

[0171] Regarding the weight of blood donors, the normal weight range is usually 30-300 kg. For those whose weight exceeds the normal range, if the abnormal weight data exceeds three digits and the first digit is less than 3, the first three digits can be retained as the actual weight. For example, if the abnormal weight recorded is 1129, it can be adjusted to 112, because 1129 kg is an unreasonable weight. If the abnormal weight data is three digits, if the first digit is greater than or equal to 3, the first two digits will be retained. For example, if the abnormal weight recorded is 366, it can be adjusted to 36.

[0172] The blood donor identity and education level in the blood donation record data are categorized and merged separately.

[0173] Typically, blood donor identity and education level are categorical variables with a high cardinality. Unlike variables like gender, which have only two categories (male or female), these variables can have dozens or even more categories. Categorical variables with a high cardinality can be merged. This means combining variables with many categories into a smaller number.

[0174] According to the blood donation location of the blood donor, data is derived to generate the blood donation point category and jurisdiction information, and the blood donation point category and jurisdiction information are added to the blood donation record data.

[0175] Based on the blood donation location information, derived variables such as blood donation site category and jurisdiction can also be derived. For example, assuming that the blood donation site is a shopping mall, the blood donation site category can be derived as "shopping mall" and the jurisdiction can be the jurisdiction where the shopping mall is located.

[0176] In addition, for other information, the very few variables with "unspecified" gender, null blood donation response, and null identity can be deleted.

[0177] In actual applications, the types of data contained in the blood donor record data may not be limited to the above types, and the data preprocessing methods are also not limited to the above types.

[0178] In summary, the blood donation return prediction method provided in this embodiment includes: obtaining a blood donor data set, the blood donor data set including at least one blood donation record data of each blood donor; constructing blood donation survival analysis data of each blood donor based on the blood donor data set; generating blood donation return estimation results corresponding to each observation time point based on the blood donation survival analysis data of each blood donor; and determining at least one blood donation return analysis result based on the blood donation return estimation results corresponding to each observation time point. This method collects blood donation record data from each blood donor during the historical study period to form a blood donation data set, and then constructs survival analysis data for each blood donor based on the blood donation record data of each blood donor; by classifying each blood donor into blood donation groups, and based on the survival analysis data of each blood donor in each blood donation group, generates blood donation return estimation results corresponding to different observation time points. The blood donation return estimation results represent the average number of blood donation returns of blood donors at the observation time point. Based on the blood donation return estimation results, the blood donation return behavior of different blood donor groups at each observation time point can be analyzed, thereby analyzing blood donor groups with high blood donation potential for recall guidance in subsequent blood donation recalls. Among them, since each blood donor is a user who actually donated blood during the historical study period, and the blood donation record data of the blood donor is also real record data generated based on the blood donor's blood donation behavior, the blood donation return behavior of real blood donors can be analyzed based on real historical data, and effective blood donation return analysis results can be obtained.

[0179] Figure 14 A schematic diagram of an electronic device provided in an embodiment of the present application is shown in FIG. Figure 14 As shown, an automated service and an automated service calling interface are deployed on the electronic device; in response to the input blood donation record data of the blood donor, the automated service is called through the calling interface; the automated service is used to execute the blood donation return prediction method in the above embodiment during operation.

[0180] This embodiment can construct the complete processing flow of the above-mentioned blood donation return prediction method into an automated service, deploy the automated service on an electronic device, build a service call interface based on the automated service, and then build an interactive network application.

[0181] Users can upload blood donation records through the automated service's application interface and call the automated service's call interface to run the automated service. Following the aforementioned blood donation return prediction process, the system implements a series of processes, including predicting blood donation returns for groups of donors, predicting blood donation returns for individual donors, fitting and training the model, and generating blood donation return analysis curves. The system also provides a download function for generated results.

[0182] The following describes the devices, equipment, storage media, etc. used to execute the blood donation return prediction method provided in this application. The specific implementation process and technical effects are described above and will not be repeated below.

[0183] Figure 15 This is a schematic diagram of a blood donation return prediction device provided in an embodiment of the present application. The functions implemented by the blood donation return prediction device correspond to the steps performed by the above method. The device can be understood as the above electronic device or server, or the processor of the server, or it can also be understood as a component independent of the above server or processor that implements the functions of the present application under the control of the server, such as Figure 15 As shown, the device may include: an acquisition module 150, a data construction module 151, a generation module 152 and a determination module 153.

[0184] An acquisition module 150 is configured to acquire a blood donor data set, wherein the blood donor data set includes at least one blood donation record data of each blood donor, and the blood donation record data includes basic information and blood donation related information; A data construction module 151 is used to construct blood donation survival analysis data for each blood donor based on the blood donor data set. The blood donation survival analysis data includes: survival analysis variables and repeated event survival analysis variables. The survival analysis variables are composed of event occurrence time and event flags. The repeated event survival analysis variables are composed of start time, end time and event flags. The generating module 152 is used to generate the blood donation return estimation results corresponding to each observation time point based on the blood donation survival analysis data of each blood donor; the blood donation return estimation results include: the average number of blood donation returns; The determination module 153 is used to determine at least one blood donation return analysis result based on the blood donation return estimation result corresponding to each observation time point.

[0185] Optionally, the data construction module 151 is specifically configured to sort each blood donation record data of each blood donor according to the blood donation time in each blood donation record data of each blood donor, and generate a sorting result of each blood donation record of each blood donor; According to the blood donation record sorting results of the blood donor, the event occurrence time and event flag corresponding to each blood donation record data of the blood donor are determined; the event occurrence time is used to indicate the interval between two adjacent blood donations; the event flag is used to indicate whether a blood donation return behavior occurs; According to the event occurrence time and event flag corresponding to each blood donation record data, the survival analysis variables corresponding to each blood donation record data of the blood donor are constructed; According to the sorting results of the blood donation records of the blood donors and the event occurrence time and event flag corresponding to each blood donation record data of the blood donors, the repeated event survival analysis variables corresponding to each blood donation record data of the blood donors are constructed.

[0186] Optionally, the data construction module 151 is specifically configured to determine whether the next blood donation record exists in the current blood donation record data according to the blood donation record sorting result of the blood donor; If yes, then determine the time of occurrence of the event corresponding to the current blood donation record data according to the blood donation time in the current blood donation record data and the blood donation time in the next blood donation record data, and determine the event flag corresponding to the current blood donation record data as the first flag, which indicates that the blood donation return behavior has occurred; If not, the event occurrence time corresponding to the current blood donation record data is determined based on the blood donation return analysis end time and the blood donation time in the current blood donation record data, and the event flag corresponding to the current blood donation record data is determined as the second flag, and the second flag indicates that a missing event has occurred.

[0187] Optionally, the data construction module 151 is specifically used to determine the start time and end time corresponding to each blood donation record data according to the blood donation record sorting result of the blood donor and the event occurrence time corresponding to each blood donation record data of the blood donor; the start time is used to indicate the time when the blood donor may donate blood next time based on the time of the first blood donation; the end time is used to indicate the time when the blood donor returns to donate blood or the time when the blood donation return analysis ends; According to the start time and end time corresponding to each blood donation record data and the event flag corresponding to each blood donation record data, a repeated event survival analysis variable corresponding to each blood donation record data of the blood donor is constructed.

[0188] Optionally, the data construction module 151 is specifically configured to calculate, for the current blood donation record data, the cumulative time of the event occurrence times corresponding to the current blood donation record data and all blood donation record data before the current blood donation record data; The accumulated time is used as the end time corresponding to the current blood donation record data; The end time of the previous blood donation record data corresponding to the current blood donation record data is used as the start time corresponding to the current blood donation record data.

[0189] Optionally, the generating module 152 is specifically configured to classify each blood donor into a group according to at least one group classification method, and determine at least one blood donor group; Based on the blood donation survival analysis data of each blood donor in each blood donor group, the blood donation return estimation of each blood donor group is performed to generate the blood donation return estimation results corresponding to each blood donor group at each observation time point.

[0190] Optionally, the generating module 152 is specifically configured to sort the blood donation record data in the blood donor group according to the end time corresponding to each blood donation record data of each blood donor in the blood donor group, to obtain a sorting result of the blood donation records of the blood donor group; According to the results of the blood donation record sorting of the blood donor group, the number of designated blood donors corresponding to each observation time point was determined. The designated blood donors were the blood donors who had not returned to donate blood and were under observation until each observation time point. Each observation time point corresponded to one day in the study period. According to the number of designated blood donors corresponding to each observation time point, the blood donation return estimate corresponding to each observation time point is determined; the blood donation return estimate is used to represent the average number of blood donation return events; Determine the upper and lower limits of the estimated blood donation return value corresponding to each observation time point based on the number of designated blood donors corresponding to each observation time point, the event flag corresponding to each blood donation record data at each observation time point, and the estimated blood donation return value at each observation time point; The estimated blood donation return value corresponding to each observation time point and the upper limit and lower limit of the estimated blood donation return value corresponding to each observation time point are used as the estimated blood donation return results corresponding to each observation time point of the blood donor group.

[0191] Optionally, the generating module 152 is specifically configured to determine each piece of blood donation record data included up to the current observation time point according to the blood donation record sorting result of the blood donor group; According to the event flags corresponding to each blood donation record data and the designated blood donor calculation strategy corresponding to the event flags, the number of designated blood donors under each blood donation record data is calculated in turn, and the number of designated blood donors under the last blood donation record data is used as the number of designated blood donors corresponding to the current observation time point.

[0192] Optionally, the generation module 152 is specifically used to determine the number of designated blood donors under the current blood donation record data based on the event flag corresponding to the current blood donation record data and the number of designated blood donors under the previous blood donation record data corresponding to the current blood donation record data, using the calculation strategy corresponding to the event flag, until the number of designated blood donors under the last blood donation record data is calculated.

[0193] Optionally, the generation module 152 is specifically used to determine the blood donation return estimate corresponding to the current observation time point based on the number of designated blood donors corresponding to the current observation time point and the blood donation return estimate at the previous observation time point corresponding to the current observation time point.

[0194] Optionally, the generating module 152 is specifically configured to determine the data identifier of the blood donor corresponding to each piece of blood donation record data according to the event flag corresponding to each piece of blood donation record data at the current observation time point, wherein the data identifier is used to indicate whether the blood donor has a blood donation return event at the current observation time point; Determine the estimated variance of the blood donation return corresponding to the current observation time point based on the estimated variance of the blood donation return at the previous observation time point corresponding to the current observation time point, the number of designated blood donors corresponding to the current observation time point, and the data identifiers of the blood donors corresponding to each blood donation record data at the current observation time point; According to the blood donation return estimation variance corresponding to the current observation time point and the blood donation return estimation value at the current observation time point, the upper limit and the lower limit of the blood donation return estimation value corresponding to the current observation time point are determined respectively.

[0195] Optionally, the determination module 153 is specifically configured to draw a blood donation return analysis curve corresponding to each blood donor group based on the blood donation return estimate value corresponding to each observation time point and the upper limit and lower limit of the blood donation return estimate value corresponding to each observation time point, wherein the blood donation return analysis curve is used to show the mapping relationship between the observation time point and the blood donation return estimate value; According to the blood donation return analysis curve corresponding to each blood donor group, the change information of the average blood donation return times of each blood donor group over time and the blood donor group with high blood donation return potential within the preset time period are determined.

[0196] Optionally, it further includes: a training module; A training module is used to construct training sample data based on each blood donation record data of each blood donor in the blood donor data set and the survival analysis variables corresponding to each blood donation record data; The training sample data is used to fit the model and train an individual blood donation return prediction model; the individual blood donation return prediction model is used to determine the blood donation return probability and the cumulative blood donation return probability of the blood donor based on the blood donation record data of the blood donor.

[0197] Optionally, the basic information of the blood donor includes at least: the gender, weight, age, identity and education level of the blood donor; the blood donation related information includes at least: time of blood donation, hemoglobin test value, blood donation reaction, blood donation amount and blood donation location; Based on the blood donor data set, before constructing the blood donation survival analysis data of each blood donor, the following steps are included: If the donor's age in the target blood donation record data is abnormal, the donor's age in the target blood donation record is corrected based on the donor's age in other blood donation record data; or the donor's age in the target blood donation record is corrected based on the age average of the category to which the donor belongs; If the blood donor's weight is abnormal in the blood donation record data, the abnormal blood donor's weight will be corrected according to the number of digits and the size of the first digit of the abnormal blood donor's weight; The blood donor identity and education level in the blood donation record data are categorized and merged separately; According to the blood donation location of the blood donor, data is derived to generate the blood donation point category and jurisdiction information, and the blood donation point category and jurisdiction information are added to the blood donation record data.

[0198] The above-mentioned device is used to execute the method provided in the above-mentioned embodiment. Its implementation principle and technical effect are similar and will not be repeated here.

[0199] The above modules can be one or more integrated circuits configured to implement the above methods, such as one or more application-specific integrated circuits (ASICs), one or more digital singular processors (DSPs), or one or more field programmable gate arrays (FPGAs). For example, when a module is implemented by scheduling program code through a processing element, the processing element can be a general-purpose processor, such as a central processing unit (CPU) or other processor that can call program code. For another example, these modules can be integrated together and implemented in the form of a system-on-a-chip (SOC).

[0200] The above modules can be connected or communicate with each other via a wired connection or a wireless connection. The wired connection may include a metal cable, an optical cable, a hybrid cable, etc., or any combination thereof. The wireless connection may include a connection in the form of a LAN, a WAN, Bluetooth, ZigBee, or NFC, or any combination thereof. Two or more modules can be combined into a single module, and any module can be divided into two or more units. Those skilled in the art will clearly understand that for the convenience and brevity of description, the specific working process of the system and device described above can refer to the corresponding process in the method embodiment, and will not be repeated in this application.

[0201] Figure 16A schematic diagram of another electronic device provided in an embodiment of the present application, which may be a computing device with data processing capabilities.

[0202] The device includes: a processor 801 and a storage medium 802 .

[0203] The storage medium 802 is used to store programs, and the processor 801 calls the programs stored in the storage medium 802 to execute the above method embodiment. The specific implementation methods and technical effects are similar and will not be repeated here.

[0204] Among them, the storage medium 802 stores program code, and when the program code is executed by the processor 801, the processor 801 executes the various steps of the blood donation return prediction method according to various exemplary embodiments of the present application described in the above "Exemplary Method" section of this specification.

[0205] The processor 801 can be a general-purpose processor, such as a central processing unit (CPU), a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components, and can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of this application. A general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly implemented as being executed by a hardware processor, or can be executed by a combination of hardware and software modules in the processor.

[0206] The storage medium 802 is a non-volatile computer-readable storage medium that can be used to store non-volatile software programs, non-volatile computer executable programs, and modules. The storage medium may include at least one type of storage medium, such as flash memory, a hard disk, a multimedia card, a card-type storage medium, a random access memory (RAM), a static random access memory (SRAM), a programmable read-only memory (PROM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a magnetic storage medium, a magnetic disk, an optical disk, and the like. The storage medium is any other medium that can be used to carry or store desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited to these. The storage medium 802 in the embodiments of the present application may also be a circuit or any other device capable of performing a storage function, used to store program instructions and / or data.

[0207] Optionally, the present application also provides a program product, such as a computer-readable storage medium, comprising a program, which is used to perform the above method embodiment when executed by a processor.

[0208] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0209] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0210] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0211] The above-mentioned integrated unit implemented in the form of a software functional unit can be stored in a computer-readable storage medium. The above-mentioned software functional unit is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) or a processor (English: processor) to perform some steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: USB flash drives, mobile hard drives, read-only storage media (English: Read-Only Memory, abbreviated: ROM), random access storage media (English: Random Access Memory, abbreviated: RAM), magnetic disks or optical disks, and other media that can store program code.

Claims

1. A blood donation return prediction method, characterized in that: include: Acquire a blood donor data set, wherein the blood donor data set includes at least one blood donation record data of each blood donor, and the blood donation record data includes: basic information and blood donation related information; Constructing blood donation survival analysis data for each blood donor based on the blood donor data set, wherein the blood donation survival analysis data includes: a survival analysis variable and a repeated event survival analysis variable, wherein the survival analysis variable is composed of an event occurrence time and an event flag, and the repeated event survival analysis variable is composed of a start time, an end time, and an event flag; Generate blood donation return estimation results corresponding to each observation time point based on the blood donation survival analysis data of each blood donor; the blood donation return estimation results include: average number of blood donation returns; At least one blood donation return analysis result is determined based on the blood donation return estimation results corresponding to each observation time point.

2. The method according to claim 1, characterized in that The step of constructing blood donation survival analysis data of each blood donor based on the blood donor data set includes: sorting the blood donation records of each blood donor according to the blood donation time in each blood donation record data of each blood donor to generate a blood donation record sorting result of each blood donor; Determine, based on the sorting results of the blood donation records of the blood donor, the event occurrence time and event flag corresponding to each piece of blood donation record data of the blood donor; the event occurrence time is used to indicate the interval between two adjacent blood donations; the event flag is used to indicate whether a blood donation return behavior occurs; constructing a survival analysis variable corresponding to each piece of blood donation record data of the blood donor according to the event occurrence time and event flag corresponding to each piece of blood donation record data; According to the sorting results of the blood donation records of the blood donors and the event occurrence time and event flag corresponding to each piece of blood donation record data of the blood donors, repeated event survival analysis variables corresponding to each piece of blood donation record data of the blood donors are constructed.

3. The method according to claim 2, characterized in that The step of determining the event occurrence time and event flag corresponding to each piece of blood donation record data of the blood donor according to the blood donation record sorting result of the blood donor includes: According to the sorting result of the blood donation records of the blood donor, determining whether the current blood donation record data contains the next blood donation record data; If so, determining the time of occurrence of the event corresponding to the current blood donation record data according to the blood donation time in the current blood donation record data and the blood donation time in the next blood donation record data, and determining the event flag corresponding to the current blood donation record data as the first flag, wherein the first flag indicates that a blood donation return behavior has occurred; If not, the event occurrence time corresponding to the current blood donation record data is determined based on the blood donation return analysis end time and the blood donation time in the current blood donation record data, and the event flag corresponding to the current blood donation record data is determined to be the second flag, which represents the occurrence of a missing event.

4. The method according to claim 2, characterized in that The method of constructing a repeated event survival analysis variable corresponding to each blood donation record data of the blood donor according to the blood donation record sorting result of the blood donor and the event occurrence time and event flag corresponding to each blood donation record data of the blood donor includes: Determine the start time and end time corresponding to each piece of blood donation record data based on the blood donation record sorting result of the blood donor and the time of occurrence of the event corresponding to each piece of blood donation record data of the blood donor; the start time is used to indicate the time when the blood donor may donate blood next time based on the time of the first blood donation; the end time is used to indicate the time when the blood donor returns to donate blood or the time when the blood donation return analysis ends; According to the start time and end time corresponding to each piece of blood donation record data and the event flag corresponding to each piece of blood donation record data, a repeated event survival analysis variable corresponding to each piece of blood donation record data of the blood donor is constructed.

5. The method according to claim 4, characterized in that The step of determining the start time and end time corresponding to each piece of blood donation record data according to the blood donation record sorting result of the blood donor and the event occurrence time corresponding to each piece of blood donation record data of the blood donor comprises: For the current blood donation record data, calculate the cumulative time of the event occurrence time corresponding to the current blood donation record data and all blood donation record data before the current blood donation record data; Using the accumulated time as the end time corresponding to the current blood donation record data; The end time of the previous blood donation record data corresponding to the current blood donation record data is used as the start time corresponding to the current blood donation record data.

6. The method according to claim 2, characterized in that The method of generating blood donation return estimation results corresponding to each observation time point based on the blood donation survival analysis data of each blood donor includes: Classifying each blood donor into a group according to at least one group classification method to determine at least one blood donor group; Based on the blood donation survival analysis data of each blood donor in each blood donor group, the blood donation return estimation of each blood donor group is performed to generate the blood donation return estimation results corresponding to each blood donor group at each observation time point.

7. The method according to claim 4, characterized in that The method of performing blood donation return estimation on each blood donor group based on the blood donation survival analysis data of each blood donor in each blood donor group, and generating blood donation return estimation results corresponding to each blood donor group at each observation time point, includes: sorting the blood donation record data of the blood donor group according to the end time corresponding to each blood donation record data of each blood donor in the blood donor group to obtain a blood donation record sorting result of the blood donor group; Determining the number of designated blood donors corresponding to each observation time point based on the blood donation record sorting results of the blood donor group, wherein the designated blood donors are blood donors who have not returned to donate blood and are under observation until each observation time point; each observation time point corresponds to a day within the study period; Determine the blood donation return estimate corresponding to each observation time point based on the number of designated blood donors corresponding to each observation time point; the blood donation return estimate is used to represent the average number of occurrences of blood donation return events; Determine the upper and lower limits of the estimated blood donation return value corresponding to each observation time point based on the number of designated blood donors corresponding to each observation time point, the event flag corresponding to each blood donation record data at each observation time point, and the estimated blood donation return value at each observation time point; The blood donation return estimation value corresponding to each observation time point and the upper limit and lower limit of the blood donation return estimation value corresponding to each observation time point are used as the blood donation return estimation results corresponding to each observation time point of the blood donor group.

8. The method according to claim 7, characterized in that The step of determining the number of designated blood donors corresponding to each observation time point according to the sorting results of the blood donation records of the blood donor group includes: Determining each piece of blood donation record data included up to a current observation time point according to the blood donation record sorting result of the blood donor group; According to the event flag corresponding to each blood donation record data and the designated blood donor calculation strategy corresponding to the event flag, the number of designated blood donors under each blood donation record data is calculated in turn, and the number of designated blood donors under the last blood donation record data is used as the number of designated blood donors corresponding to the current observation time point.

9. The method according to claim 8, characterized in that The method of sequentially calculating the number of designated blood donors under each piece of blood donation record data according to the event flag corresponding to each piece of blood donation record data and the designated blood donor calculation strategy corresponding to the event flag includes: According to the event flag corresponding to the current blood donation record data and the number of designated blood donors under the previous blood donation record data corresponding to the current blood donation record data, the calculation strategy corresponding to the event flag is adopted to determine the number of designated blood donors under the current blood donation record data until the number of designated blood donors under the last blood donation record data is calculated.

10. The method according to claim 7, characterized in that Determining the blood donation return estimate corresponding to each observation time point based on the number of designated blood donors corresponding to each observation time point includes: The estimated blood donation return value corresponding to the current observation time point is determined based on the number of designated blood donors corresponding to the current observation time point and the estimated blood donation return value at the previous observation time point corresponding to the current observation time point.

11. The method according to claim 7, characterized in that The method of determining the upper limit and lower limit of the estimated blood donation return value corresponding to each observation time point based on the number of designated blood donors corresponding to each observation time point, the event flag corresponding to each blood donation record data at each observation time point, and the estimated blood donation return value at each observation time point includes: Determine the data identifier of the blood donor corresponding to each piece of blood donation record data according to the event flag corresponding to each piece of blood donation record data at the current observation time point, wherein the data identifier is used to indicate whether the blood donor has a blood donation return event at the current observation time point; Determine the estimated variance of the blood donation return corresponding to the current observation time point based on the estimated variance of the blood donation return at the previous observation time point corresponding to the current observation time point, the number of designated blood donors corresponding to the current observation time point, and the data identifiers of the blood donors corresponding to each blood donation record data at the current observation time point; According to the blood donation return estimation variance corresponding to the current observation time point and the blood donation return estimation value at the current observation time point, the upper limit and the lower limit of the blood donation return estimation value corresponding to the current observation time point are determined respectively.

12. The method according to claim 7, characterized in that Determining at least one blood donation return analysis result based on the blood donation return estimation results corresponding to each observation time point includes: According to the blood donation return estimate value corresponding to each blood donor group at each observation time point and the upper and lower limits of the blood donation return estimate value corresponding to each observation time point, a blood donation return analysis curve corresponding to each blood donor group is drawn, and the blood donation return analysis curve is used to show the mapping relationship between the observation time point and the blood donation return estimate value; According to the blood donation return analysis curve corresponding to each blood donor group, the change information of the average blood donation return times of each blood donor group over time and the blood donor group with high blood donation return potential within the preset time period are determined.

13. The method according to claim 2, characterized in that Also includes: constructing training sample data according to each blood donation record data of each blood donor in the blood donor data set and the survival analysis variable corresponding to each blood donation record data; The training sample data is used to perform model fitting and train an individual blood donation return prediction model; the individual blood donation return prediction model is used to determine the blood donation return probability and the cumulative blood donation return probability of the blood donor based on the blood donor's first blood donation record data.

14. The method according to claim 1, wherein The basic information of the blood donor includes at least: the gender, weight, age, identity and education level of the blood donor; the blood donation related information includes at least: time of blood donation, hemoglobin test value, blood donation reaction, blood donation amount and blood donation location; Before constructing the blood donation survival analysis data of each blood donor based on the blood donor data set, the method includes: If there is an abnormality in the target blood donation record data of the blood donor, the age of the blood donor in the target blood donation record is corrected according to the age of the blood donor in other blood donation record data of the blood donor; or the age of the blood donor in the target blood donation record is corrected according to the age average of the category to which the blood donor belongs; If the blood donor's weight is abnormal in the blood donation record data, the abnormal blood donor's weight will be corrected according to the number of digits and the size of the first digit of the abnormal blood donor's weight; The blood donor identity and education level in the blood donation record data are categorized and merged separately; According to the blood donation location of the blood donor, data is derived to generate the blood donation point category and jurisdiction information, and the blood donation point category and jurisdiction information are added to the blood donation record data.

15. An electronic device, characterized in that: An automation service and a calling interface of the automation service are deployed on the electronic device; In response to the input blood donation record data of the blood donor, calling the automation service through the calling interface; The automated service is used to execute the blood donation return prediction method according to any one of claims 1 to 14 during operation.

Images