Comprehensive nursing management system based on big data analysis

Abstract

The comprehensive nursing management system based on big data analysis belongs to the technical field of artificial intelligence model management, and comprises a fairness data management unit, a fairness quantitative evaluation unit, a risk trend early warning unit and a self-adaptive intervention response unit; the fairness data management unit is used for dividing medical data nodes into a preset number of medical resource level groups, and collecting the average prediction error rate and the total population of each medical resource level group; the fairness quantitative evaluation unit is used for receiving the average prediction error rate and the total population sent by the fairness data management unit, performing fairness quantitative analysis, and generating a medical resource allocation deviation index, which ensures the continuity and efficiency of risk management.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of artificial intelligence model governance, in particular to a comprehensive nursing management system based on big data analysis. BACKGROUND

[0002] In the fairness evaluation of medical AI models, traditional methods mainly rely on manual auditing and analysis of single indicators such as overall accuracy of the model; these methods have obvious shortcomings in measuring whether the model produces structural bias for groups with different levels of medical resources, and often cannot dynamically and quantitatively reveal the distribution difference of model errors among different groups; this situation makes it difficult for managers to discover and evaluate the possible exacerbation of medical resource allocation inequality.

[0003] The above situation and shortcomings mainly result from the limitations of evaluation methods and concepts. First, relying on a single overall performance indicator, such as total accuracy, can mask poor performance of the model in certain disadvantaged groups, and cannot effectively measure structural unfairness; second, the manual auditing method has a long response cycle and lacks continuity, resulting in significant lag in identifying fairness risks; in addition, traditional methods are mostly reactive, lacking the ability to predict the trend of fairness deterioration, and can only intervene after the problem occurs and breaks through the threshold, failing to avoid risks in advance.

[0004] As a result, when the prediction error rate of the model for certain groups begins to deteriorate, decision-makers cannot obtain clear and quantitative early warning signals in a timely manner, which delays the best opportunity to take intervention measures such as adjusting the model algorithm or optimizing resource allocation, and may unintentionally exacerbate or solidify existing health inequalities in the application of medical AI systems.

[0005] The above information disclosed in the background section is only used to strengthen the understanding of the background of the present disclosure, and therefore it can include information that does not constitute prior art known to those of ordinary skill in the art. SUMMARY

[0006] The purpose of the present application is to provide a comprehensive nursing management system based on big data analysis to solve the problems raised in the background.

[0007] The technical solution of the present application is to divide the medical data nodes into a predetermined number of medical resource level groups, and for each medical resource level group, to collect its average prediction error rate and total population.

[0008] The fairness data management unit is used to divide the medical data nodes into a predetermined number of medical resource level groups, and for each medical resource level group, to collect its average prediction error rate and total population.

[0009] The fairness quantification evaluation unit is configured to receive the average prediction error rate and the total population number sent by the fairness data management unit, perform fairness quantification analysis, and generate a medical resource allocation deviation index;

[0010] The risk trend early warning unit is configured to receive the medical resource allocation deviation index of the current period sent by the fairness quantification evaluation unit, perform risk trend evaluation, and generate a fairness safety distance.

[0011] The adaptive intervention response unit is configured to receive the fairness safety distance sent by the risk trend early warning unit, compare and analyze the fairness safety distance with a preset intervention response threshold, and generate a risk intervention signal according to the comparison and analysis result.

[0012] Preferably, the fairness quantification evaluation unit performs the fairness quantification analysis and generates the medical resource allocation deviation index through the following process:

[0013] Obtain the proportion of each medical resource level group in the nodes, and set the proportion as a group weight;

[0014] Calculate a weighted average prediction error rate based on the group weight and the average prediction error rate;

[0015] Calculate a population size penalty factor based on the total population number of each medical resource level group;

[0016] Calculate the standard deviation of the average prediction error rate of each medical resource level group;

[0017] Generate the medical resource allocation deviation index based on the standard deviation, the weighted average prediction error rate, and the population size penalty factor.

[0018] Preferably, the calculation process of the weighted average prediction error rate is as follows:

[0019] Multiply the group weight of each medical resource level group by the corresponding average prediction error rate to obtain a weighted error value of each group;

[0020] Sum all the weighted error values of the medical resource level groups, and set the sum as the weighted average prediction error rate.

[0021] Preferably, the calculation process of the population size penalty factor is as follows:

[0022] Obtain the total population number of all the medical resource level groups, and calculate the standard deviation and the average value of the total population number;

[0023] Summing the ratio between the standard deviation and the average value with the numerical value 1 and setting the resulting sum value as the population size penalty factor.

[0024] Preferably, the risk trend early warning unit performs the risk trend assessment and generates the fairness safety distance in the following process:

[0025] Obtain the medical resource allocation deviation index of the current period and the medical resource allocation deviation index of the last period;

[0026] Based on the medical resource allocation deviation index of the current period, the medical resource allocation deviation index of the last period, and the preset evaluation period time interval, calculate the change rate of the fairness deviation index;

[0027] Obtain the preset fairness deviation index failure boundary threshold value, and set the difference between the failure boundary threshold value and the medical resource allocation deviation index of the current period as the basic safety distance;

[0028] Obtain the preset risk sensitivity coefficient, and based on the change rate of the fairness deviation index and the risk sensitivity coefficient, calculate the risk attenuation factor;

[0029] Multiply the basic safety distance by the risk attenuation factor, and set the resulting product as the fairness safety distance.

[0030] Preferably, the calculation process of the change rate of the fairness deviation index is as follows:

[0031] Subtract the medical resource allocation deviation index of the current period from the medical resource allocation deviation index of the last period to obtain an index difference value;

[0032] Divide the index difference value by the evaluation period time interval, and set the resulting quotient value as the change rate of the fairness deviation index.

[0033] Preferably, the adaptive intervention response unit generates the risk intervention signal in the following process:

[0034] Compare and analyze the fairness safety distance with the preset intervention response threshold value;

[0035] If the fairness safety distance is less than the preset intervention response threshold value, generate the risk intervention signal;

[0036] If the fairness safety distance is greater than or equal to the preset intervention response threshold value, do not generate any signal.

[0037] The application provides a comprehensive nursing management system based on big data analysis by improvement, and has the following improvements and advantages compared with the prior art.

[0038] 1. The system realizes systematic and automatic management of model fairness risk by constructing a complete architecture including fairness data management unit, fairness quantitative evaluation unit, risk trend early warning unit and adaptive intervention response unit. The architecture converts the abstract fairness concept into a series of precise and connected technical operations, i.e. data collection, quantitative analysis, trend evaluation and intervention response, thereby constructing a complete closed-loop control process; the passive situation of relying on manual audit and delayed response in the past is changed, and the continuity and efficiency of risk management are ensured;

[0039] 2. The medical resource allocation deviation index designed by the system can deeply reveal the structural unfairness in model prediction. Instead of using a single accuracy indicator, the system calculates the standard deviation of the average prediction error rate of each medical resource level group, and compares it with the weighted average prediction error rate, thereby quantifying the relative dispersion degree of error among different groups. More importantly, the index also introduces a population size penalty factor. This design makes the sensitivity of the index significantly enhanced when the area serving a large population has prediction deviation, thus accurately solving the problem that traditional evaluation methods cannot effectively measure the structural defects of service resource allocation;

[0040] 3. The fairness safety distance generated by the system realizes the forward-looking evaluation and early warning of risk. Instead of relying solely on the current absolute value of the medical resource allocation deviation index, the system innovatively introduces the change rate of the index as a key consideration factor. When the fairness index shows an accelerating deterioration trend, even if its absolute value has not reached the failure boundary, the fairness safety distance will also shrink sharply. This design enables the system to trigger an early warning before the risk actually occurs and breaks through the threshold, thereby gaining valuable time for taking intervention measures and changing the limitations of traditional methods that can only respond passively after the fact;

[0041] 4. The adaptive intervention response mechanism established by the system ensures the timeliness and reliability of risk avoidance measures. By continuously comparing the dynamically calculated fairness safety distance with the preset intervention response threshold, the system establishes a clear and automatic decision trigger mechanism. This mechanism determines whether to generate a risk intervention signal based solely on forward-looking risk assessment data, thereby eliminating the delay and subjectivity that may be caused by manual judgment, greatly ensuring the response efficiency and operational reliability of the entire risk avoidance system. BRIEF DESCRIPTION OF DRAWINGS

[0042] The application will be further explained in conjunction with the drawings and examples:

[0043] Figure 1 is a flow chart of the system of the present application. DETAILED DESCRIPTION

[0044] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application is further described in detail below with reference to specific embodiments.

[0045] Embodiment 1

[0046] Referring to Figure 1 The present application provides a comprehensive nursing management system based on big data analysis, which comprises a fairness data management unit, a fairness quantitative evaluation unit, a risk trend early warning unit and a self-adaptive intervention response unit.

[0047] The fairness data management unit is used to divide the medical data nodes into a preset number of medical resource level groups, and collect the average prediction error rate and the total population of each medical resource level group.

[0048] The fairness quantitative evaluation unit is used to receive the average prediction error rate and the total population sent by the fairness data management unit, perform fairness quantitative analysis, and generate a medical resource allocation deviation index.

[0049] The risk trend early warning unit is used to receive the medical resource allocation deviation index of the current period sent by the fairness quantitative evaluation unit, perform risk trend evaluation, and generate a fairness safety distance.

[0050] The self-adaptive intervention response unit is used to receive the fairness safety distance sent by the risk trend early warning unit, compare and analyze it with a preset intervention response threshold, and generate a risk intervention signal according to the comparison and analysis result.

[0051] In the embodiments of the present application, the medical data nodes can be independent medical institutions such as hospitals, community health service centers, or geographical units divided according to administrative divisions such as districts and counties. The method of dividing these nodes into medical resource level groups can be based on publicly available quantitative indicators. For example, a comprehensive scoring method can be used, which includes the following steps:

[0052] Select multiple key resource indicators, such as the number of practicing physicians per thousand people, the number of hospital beds per thousand people, and the annual per capita medical financial investment;

[0053] Non-dimensionalize the above indicators of each node;

[0054] Calculate the comprehensive resource score of each node by assigning a preset weight to each indicator, for example, 40% for the number of physicians, 30% for the number of beds, and 30% for financial investment;

[0055] According to the comprehensive resource score of all nodes, statistical methods such as K-means clustering algorithm or according to the score quantile are used to divide the nodes into three medical resource level groups: high, medium and low;

[0056] The application provides a comprehensive nursing management system based on big data analysis; a fairness data management unit divides medical data nodes into a preset number of medical resource level groups. The fairness data management unit collects the average prediction error rate and the total population of each medical resource level group;

[0057] Average prediction error rate refers to the prediction performance of a specific health prediction model in a certain medical resource level group; for example, the model can be a logistic regression model or a gradient boosting tree model for predicting the 30-day readmission risk of diabetes patients; the error rate can be calculated by using the mean absolute error or the root mean square error; when collecting data, for all medical data nodes in the group , the prediction model is run, and the predicted value is compared with the true result to calculate the average prediction error rate of the group as a whole ;

[0058] The fairness quantification evaluation unit receives the average prediction error rate and the total population sent by the fairness data management unit; the fairness quantification evaluation unit performs fairness quantification analysis and generates a medical resource allocation deviation index; the risk trend early warning unit receives the medical resource allocation deviation index of the current period sent by the fairness quantification evaluation unit; the risk trend early warning unit performs risk trend evaluation and generates a fairness safety distance; the adaptive intervention response unit receives the fairness safety distance sent by the risk trend early warning unit; the adaptive intervention response unit compares and analyzes it with the preset intervention response threshold; the adaptive intervention response unit generates a risk intervention signal according to the comparison and analysis result; the system converts the abstract fairness risk into specific data flow and intervention signal through the chain cooperation of each unit, realizes the closed-loop management from quantitative definition, trend evaluation to adaptive response, and overcomes the defects of traditional methods relying on manual audit and response lag.

[0059] Embodiment 2

[0060] The process of the fairness quantification evaluation unit performing fairness quantification analysis and generating a medical resource allocation deviation index includes:

[0061] Obtain the proportion of nodes occupied by each medical resource level group, and set it as the group weight;

[0062] Based on the group weight and the average prediction error rate, the weighted average prediction error rate is calculated;

[0063] The population size penalty factor is calculated based on the total population of each group with different levels of medical resources.

[0064] Calculate the standard deviation of the average prediction error rate for each group with different levels of medical resources;

[0065] A medical resource allocation bias index is generated based on standard deviation, weighted average prediction error rate, and population size penalty factor.

[0066] The calculation process for the weighted average prediction error rate is as follows:

[0067] The weighted error value for each group is obtained by multiplying the group weight of each medical resource level group with its corresponding average prediction error rate.

[0068] The weighted error values ​​of all medical resource level groups are summed, and the sum is set as the weighted average prediction error rate.

[0069] The calculation process for the population size penalty factor is as follows:

[0070] Obtain the total population for all groups with different healthcare resource levels, and calculate the standard deviation and mean of the total population.

[0071] The ratio between the standard deviation and the mean is summed with the value 1, and the resulting sum is set as the population size penalty factor.

[0072] This embodiment is an explanation based on Embodiment 1. Specifically, the process by which the fairness quantification assessment unit performs fairness quantification analysis and generates a medical resource allocation deviation index includes:

[0073] The fairness quantification assessment unit obtains the proportion of nodes occupied by each medical resource level group and sets it as the group weight; based on the group weight and the average prediction error rate, the fairness quantification assessment unit calculates the weighted average prediction error rate; the formula for calculating the weighted average prediction error rate is defined as:

[0074]

[0075] in This represents the weighted average prediction error rate. This indicates the number of people in a group with a pre-defined level of medical resources. Indicates the group number. Indicates the first Group weight of each group Indicates the first The average prediction error rate for each group;

[0076] The fairness quantitative evaluation unit calculates a population size penalty factor based on the total population of each medical resource level group. The calculation formula of the population size penalty factor is defined as:

[0077]

[0078] wherein denotes the population size penalty factor, denotes the standard deviation of the total population of all groups, denotes the average value of the total population of all groups; the preset population size penalty factor works in this way: when the error rate of a group with large dispersion is also significantly different from other groups in terms of population size, the final deviation index will be amplified, thereby making the system more sensitive to areas serving a large population but performing poorly;

[0079] The fairness quantitative evaluation unit calculates the standard deviation of the average prediction error rate of each medical resource level group. The fairness quantitative evaluation unit generates a medical resource allocation deviation index based on the standard deviation, the weighted average prediction error rate, and the population size penalty factor. The calculation formula of the medical resource allocation deviation index is defined as:

[0080]

[0081] wherein denotes the medical resource allocation deviation index, denotes the standard deviation of the average prediction error rate of each group; this medical resource allocation deviation index quantifies the relative dispersion degree of errors among different groups by introducing the ratio of the standard deviation to the weighted average error rate, and is weighted by the population size penalty factor, so that the evaluation result can accurately reflect the severity of the prediction deviation in areas serving a large population, solving the problem that the traditional single accuracy index cannot measure structural injustice.

[0082] Embodiment 3

[0083] The process of the risk trend early warning unit for risk trend evaluation and generation of the fairness safety distance includes:

[0084] Obtaining the medical resource allocation deviation index of the current period and the medical resource allocation deviation index of the previous period;

[0085] Based on the medical resource allocation deviation index of the current period, the medical resource allocation deviation index of the previous period, and the preset evaluation period time interval, the change rate of the fairness deviation index is calculated;

[0086] The preset evaluation period time interval The choice of monitoring model should be matched with the characteristics of the healthcare problem being monitored; for rapidly changing scenarios, such as seasonal influenza or acute infectious diseases, the model should predict fairness in monitoring. It can be set to a shorter period, such as one week; for scenarios with relatively stable changes, such as resource allocation models for chronic disease management effectiveness or regional health planning, It can be set to a longer period, such as one month or one quarter. In this embodiment, we set... For one week;

[0087] Obtain the preset fairness deviation index failure boundary threshold, and set the difference between the failure boundary threshold and the medical resource allocation deviation index of the current period as the basic safety distance;

[0088] Preset fairness deviation index failure boundary threshold This is the key risk boundary, and its value can be set according to one or more of the following methods:

[0089] Historical data method: Analyze the medical resource allocation deviation index over the past few years, such as 3-5 years. Historical data, with its historical mean plus two or three standard deviations as the value. ;

[0090] Policy, regulations, and laws: Referencing relevant fairness guidelines or regulatory requirements issued by the government or health authorities, quantify and set them as follows. ;

[0091] Expert consultation method: Organize public health, medical management and ethics experts to conduct an assessment and jointly agree on a threshold that represents the risk threshold;

[0092] For example, if calculated using historical data methods The historical mean is 0.15 and the standard deviation is 0.05, so we can... Set as ;

[0093] Obtain the preset risk sensitivity coefficient, and calculate the risk decay factor based on the rate of change of the fairness deviation index and the risk sensitivity coefficient;

[0094] Preset risk sensitivity coefficient Used to adjust fair safety distance Rate of change of deviation index The degree of sensitivity; The larger the value, the greater the safe distance becomes as fairness deteriorates more rapidly. The faster the contraction speed, the more sensitive the system's early warning; its value can be determined through simulation testing: input a series of historical or hypothetical data. Data, adjustment Choose a value, for example, between 0.5 and 10, and observe. The change curve should be selected to provide timely warnings while avoiding frequent false alarms due to normal fluctuations; to ensure the accuracy of the formula's dimensions, It is a parameter with a time dimension, and its unit should be the time interval of the evaluation period. The units should be consistent, for example, if If it is one week, then The unit is also week; for example, it can be set to... Zhou; For areas requiring high vigilance, a larger [scale / resource] can be selected. Value, such as ;

[0095] Multiply the base safety distance by the risk attenuation factor, and set the product as the fairness safety distance.

[0096] The calculation process for the rate of change of the fairness deviation index is as follows:

[0097] The difference between the medical resource allocation deviation index of the current period and the medical resource allocation deviation index of the previous period is obtained.

[0098] Divide the index difference by the evaluation period interval and set the resulting quotient as the rate of change of the fairness deviation index.

[0099] This embodiment is an explanation of Embodiment 2. Specifically, the process by which the risk trend early warning unit assesses risk trends and generates a fairness safety distance includes:

[0100] The risk trend early warning unit obtains the medical resource allocation deviation index for the current period and the medical resource allocation deviation index for the previous period. Based on the medical resource allocation deviation index for the current period, the medical resource allocation deviation index for the previous period, and the preset assessment period time interval, the risk trend early warning unit calculates the rate of change of the fairness deviation index. The formula for calculating the rate of change of the fairness deviation index is defined as follows:

[0101]

[0102] in This represents the rate of change of the fairness deviation index. This represents the medical resource allocation deviation index for the current period. This represents the medical resource allocation deviation index for the previous period. This indicates the preset evaluation cycle time interval.

[0103] The risk trend early warning unit acquires a preset fairness deviation index failure boundary threshold value, and sets a difference between the failure boundary threshold value and a medical resource allocation deviation index of a current period as a basic safety distance. The risk trend early warning unit acquires a preset risk sensitivity coefficient. The risk trend early warning unit calculates a risk attenuation factor based on a change rate of the fairness deviation index and the risk sensitivity coefficient, and sets a product of the basic safety distance and the risk attenuation factor as a fairness safety distance; a calculation formula of the fairness safety distance is defined as:

[0104]

[0105] wherein represents the fairness safety distance, represents the preset fairness deviation index failure boundary threshold value, represents the preset risk sensitivity coefficient; the preset fairness deviation index failure boundary threshold value is a parameter representing a maximum degree of unfairness that is tolerable by society; : this function is used to take a larger value between 0 and the change rate of the fairness deviation index; : represents the change rate of the fairness deviation index; the preset risk sensitivity coefficient is used to control a reaction speed of the system to a fairness deterioration trend; generation of the fairness safety distance realizes a forward-looking assessment of the risk by introducing the change rate and the exponential attenuation factor; when the fairness index deteriorates at an accelerated rate, even if an absolute value thereof is still within the threshold value, the safety distance is sharply reduced, so that an early warning is triggered before the risk actually occurs, thereby changing a limitation that a traditional method can only respond passively after a threshold value is broken.

[0106] Embodiment 4

[0107] The adaptive intervention response unit generates a risk intervention signal in the following process:

[0108] The fairness safety distance is compared and analyzed with a preset intervention response threshold value;

[0109] The preset intervention response threshold value is a decision boundary for starting an intervention measure, and the threshold value should be matched with a dimension and a dynamic range of the fairness safety distance , and can be set as a multi-level threshold value to realize a graded response, for example, two threshold values can be set:

[0110] A first early warning threshold value: when the initial value thereof is first lower than the initial value, that is, when the fairness safety distance is 50% of the fairness safety distance when the fairness safety distance is generated, the system triggers a yellow early warning to prompt a manager to pay attention;

[0111] A second intervention threshold value: when When the fairness safety distance is less than 20% of the initial value, the system determines that the situation is urgent, the threshold at this time is the intervention response threshold, the system generates a risk intervention signal, and requires taking substantive intervention measures;

[0112] The hierarchical setting makes the response mechanism more flexible and practical;

[0113] If the fairness safety distance is less than the preset intervention response threshold, a risk intervention signal is generated.

[0114] If the fairness safety distance is greater than or equal to the preset intervention response threshold, no signal is generated.

[0115] The embodiment is an explanation and description in embodiment 3, specifically, the process of generating a risk intervention signal by the adaptive intervention response unit includes:

[0116] The adaptive intervention response unit compares and analyzes the fairness safety distance with the preset intervention response threshold, if the fairness safety distance is less than the preset intervention response threshold, the adaptive intervention response unit generates a risk intervention signal, if the fairness safety distance is greater than or equal to the preset intervention response threshold, the adaptive intervention response unit does not generate any signal, the preset intervention response threshold is an adjustable parameter for defining the tolerance of system triggering intervention, the adaptive intervention response unit compares the dynamically calculated fairness safety distance with the static intervention response threshold, establishes an explicit and automatic decision triggering mechanism, the mechanism ensures that the intervention measures are started in time on the basis of the prospective risk assessment, avoids the delay and subjectivity caused by manual judgment, and guarantees the response efficiency and reliability of the whole risk avoidance system.

[0117] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present application but not limit the present application, although the present application is described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or replaced equivalently without departing from the spirit and scope of the technical solutions of the present application.

Claims

1. A comprehensive nursing management system based on big data analysis, characterized by, The fairness data management unit, the fairness quantification evaluation unit, the risk trend early warning unit, and the adaptive intervention response unit; The fairness data management unit is configured to divide the medical data nodes into a preset number of medical resource level groups, and collect, for each medical resource level group, an average prediction error rate and a total population thereof; The fairness quantification evaluation unit is configured to receive the average prediction error rate and the total population sent by the fairness data management unit, perform fairness quantification analysis, and generate a medical resource allocation deviation index; The risk trend early warning unit is configured to receive the medical resource allocation deviation index of the current period sent by the fairness quantification evaluation unit, perform risk trend evaluation, and generate a fairness safety distance; The adaptive intervention response unit is configured to receive the fairness safety distance sent by the risk trend early warning unit, compare and analyze the fairness safety distance with a preset intervention response threshold, and generate a risk intervention signal according to the comparison and analysis result; The fairness quantification evaluation unit performs the fairness quantification analysis and generates the medical resource allocation deviation index by: obtaining the proportion of nodes occupied by each medical resource level group and setting the proportion as a group weight; calculating a weighted average prediction error rate based on the group weight and the average prediction error rate; calculating a population size penalty factor based on the total population of each medical resource level group; calculating the standard deviation of the average prediction error rate of each medical resource level group; generating the medical resource allocation deviation index based on the standard deviation, the weighted average prediction error rate, and the population size penalty factor; The calculation process of the population size penalty factor is: obtaining the total population of all medical resource level groups and calculating the standard deviation and the average value of the total population; summing the ratio between the standard deviation and the average value and the value 1, and setting the obtained sum value as the population size penalty factor; The risk trend early warning unit performs the risk trend evaluation and generates the fairness safety distance by: obtaining the medical resource allocation deviation index of the current period and the medical resource allocation deviation index of the previous period; calculating a fairness deviation index change rate based on the medical resource allocation deviation index of the current period, the medical resource allocation deviation index of the previous period, and a preset evaluation period time interval; obtaining a preset fairness deviation index failure boundary threshold, and setting the difference between the failure boundary threshold and the medical resource allocation deviation index of the current period as a basic safety distance; obtaining a preset risk sensitivity coefficient, and calculating a risk attenuation factor based on the fairness deviation index change rate and the risk sensitivity coefficient; multiplying the basic safety distance by the risk attenuation factor, and setting the obtained product as the fairness safety distance.

2. The big data analysis-based total care management system of claim 1, wherein, The calculation process of the weighted average prediction error rate is: The group weight of each medical resource level group is multiplied by the corresponding average prediction error rate to obtain a weighted error value of each group; The weighted error values of all medical resource level groups are summed, and the obtained sum value is set as the weighted average prediction error rate. 3.The overall care management system based on big data analysis according to claim 1, wherein, The calculation process of the change rate of the fairness deviation index is: The medical resource allocation deviation index of the current period is subtracted from the medical resource allocation deviation index of the last period to obtain an index difference value; The index difference value is divided by the evaluation period time interval, and the obtained quotient value is set as the change rate of the fairness deviation index. 4.The overall care management system based on big data analysis according to claim 1, wherein, The process in which the adaptive intervention response unit generates the risk intervention signal is: The fairness safety distance is compared and analyzed with the preset intervention response threshold value; If the fairness safety distance is less than the preset intervention response threshold value, the risk intervention signal is generated; If the fairness safety distance is greater than or equal to the preset intervention response threshold value, no signal is generated.

Images