Apparatus, method and storage medium for predicting or evaluating time-dependent parameters

Abstract

The invention relates to a device, a method and a storage medium for predicting or evaluating time-related parameters. The device comprises a processor and a memory, wherein the memory is used for storing executable instructions of the processor; wherein the processor is configured to execute the executable instruction to implement the following steps: obtaining parameter sample data; determining or acquiring a distribution mode of the parameters on each time value; determining or acquiring a change mode of the parameters in time; determining or acquiring constraint conditions; based on the parameter sample data, the distribution mode, the change mode and a constraint condition, determining a time-dependent probability function, the constraint condition comprising a value range of at least one of a parameter value and time in the time-dependent probability function; a trend of variation of the parameter at a future time is predicted and / or a specific parameter value of the parameter is evaluated based on a time-dependent probability function.

Description

Background Art

[0002] In human production and life, there are many parameters that are related to time or that change with time. Such parameters include, for example, physiological / pathological parameters of the human body that depend on time, especially physiological / pathological parameters that change with time or with the growth of the human body, such as the axial length of the human eye, corneal curvature, etc. For such parameters, it may be necessary to perform statistics on the existing parameters in order to achieve a prediction of future development or an evaluation of specific parameter values. Of course, such parameters may also include physical or chemical parameters of mechanical or electronic devices. For example, aging parameters or fatigue parameters of mechanical parts or electronic devices.

[0003] In the prior art, the so-called percentile curve is usually used to predict or evaluate time-related parameters. The percentile generally represents the percentage of the number of parameters with a certain parameter value and less than the parameter value to the total number of parameters counted. For example, Figure 1 Taking the axial length of children of different ages as an example, a percentile curve is shown. This figure is excerpted from "Expert Consensus on Hyperopia Reserve, Axial Length, Corneal Curvature Reference Range and Related Genetic Factors of Chinese School-Age Children (2022)" (Public Health Ophthalmology Branch of the Chinese Preventive Medicine Association, Chinese Journal of Ophthalmology, Volume 58, Issue 2, February 2022, Pages 96-102). The horizontal axis represents the age of the child, and the vertical axis represents the length of the child's axial length. In this figure, for example, curve P3 shows that for children of different ages, the percentage of people with an axial length less than or equal to the corresponding ordinate coordinate is 3%. For example, point A in the figure indicates that for 8-year-old children, the statistical results show that the percentage of people with an axial length less than or equal to about 21.2 mm is 3%.

[0004] Although in practice this percentile curve can be used to predict the trend of parameter changes or evaluate specific parameter values, it may be inaccurate or even unreasonable due to the unsatisfactory sample data. Figure 1 The P3 curve has a significant zigzag shape which may lead to inaccurate predictions or assessments.

[0005] like Figure 1 This percentile curve of can also be smoothed by fitting (e.g., polynomial fitting). For example, Figure 1 The percentile curve shown can be generated by fitting a quadratic polynomial as shown below. Figure 2 The smoothed percentile curve shown (see Figure 2 ). However, even the fitted curve may be inaccurate, and the percentile curve may even deviate from the common sense of the parameter. Figure 2The curve shown deviates from the general knowledge of the axial length of children's eyes. For example, Figure 2 as can be seen from P10 and P25 in Figure 2 , the percentile curves even intersect. This intersection may be caused by insufficient data or the singularity of individual data, but it clearly does not conform to the general knowledge of this parameter. Moreover, when the parameter value or the corresponding time of the parameter exceeds the range of the sample data, extrapolation needs to be performed on the obtained percentile curve. For example, for the statistical results in Figure 2 , if it is desired to predict the axial length development of children at the age of 16, extrapolation of the curve is required. This extrapolation can be achieved, for example, by means of fitting estimation of the original sample data. For the application of the axial length of children's eyes, the result of polynomial fitting has been proven to be ineffective in extrapolation. In addition, since the collection of original data must be limited, and the limitations of this method itself may lead to prediction or evaluation failure (<P0 or >P100) when the parameter exceeds the range of the original data set, see Figure 2 . SUMMARY OF THE INVENTION

[0006] To solve the above problems and deficiencies in the prior art, the present invention provides a device for predicting or evaluating a time-related parameter, a method for predicting or evaluating a time-related parameter, and a computer storage medium.

[0007] According to one aspect of the present invention, there is provided a device for predicting or evaluating a time-related parameter, the device comprising:

[0008] a processor; and

[0009] a memory for storing executable instructions of the processor;

[0010] wherein the processor is configured to execute the executable instructions to perform the following steps:

[0011] - Obtain parameter sample data, the parameter sample data including a certain number of data groups, each data group including a parameter value of the parameter and a time value corresponding to the parameter value;

[0012] - Determine or obtain the distribution pattern of the parameter at each time value;

[0013] - Determine or obtain the change pattern of the parameter over time;

[0014] - Determine or obtain constraint conditions;

[0015] -determining a time-dependent probability function based on the parameter sample data, the distribution pattern, the variation pattern, and the constraint condition, wherein the constraint condition includes a value range of at least one of a parameter value and time of the parameter in the time-dependent probability function; and

[0016] - predicting the variation trend of the parameter at a future time and / or evaluating a specific parameter value of the parameter based on the time-dependent probability function.

[0017] According to another aspect of the present invention, a method for predicting or estimating a time-related parameter is provided, comprising:

[0018] - obtaining parameter sample data, wherein the parameter sample data includes a certain number of data groups, each data group includes a parameter value of a parameter and a time value corresponding to the parameter value;

[0019] - determining or obtaining the distribution pattern of the parameter at each time value;

[0020] - determining or obtaining a temporal variation pattern of said parameter;

[0021] -determine or obtain constraints;

[0022] -determining a time-dependent probability function based on the parameter sample data, the distribution pattern, the variation pattern, and the constraint condition, wherein the constraint condition includes a value range of at least one of a parameter value and time of the parameter in the time-dependent probability function; and

[0023] - predicting the variation trend of the parameter at a future time and / or evaluating a specific parameter value of the parameter based on the time-dependent probability function.

[0024] According to another aspect of the present invention, a computer-readable storage medium is provided, on which a computer program is stored. The computer program includes executable instructions. When the executable instructions are executed by a processor, the method described above is implemented.

[0025] The device and method for predicting or evaluating time-related parameters of the present invention can overcome various problems existing in the prior art. Compared with the existing solutions, the present invention can provide more reliable and accurate prediction or evaluation possibilities. Here, by determining or obtaining the distribution pattern and its change pattern over time, it is possible to better adapt to the characteristics of the parameters themselves, thereby providing a more reasonable fitting scheme, and then obtaining accurate prediction or evaluation results. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings to further illustrate features and advantages of the present invention.

[0027] Figure 1 An exemplary percentile graph of children's axial length without smoothing according to the prior art.

[0028] Figure 2 An exemplary percentile graph of smoothed axial length of children according to the prior art.

[0029] Figure 3 It is a schematic flow chart of a method according to an embodiment of the present invention.

[0030] Figure 4 Schematic graph of a time-dependent probability density function obtained by a method according to an embodiment of the present invention.

[0031] Figure 5 Schematic graph of a cumulative distribution function dependent on time obtained by a method according to an embodiment of the present invention.

[0032] Figure 6 Schematic diagram of a class percentile curve obtained by a method according to an embodiment of the present invention.

[0033] Figure 7 FIG. 4 is a schematic structural block diagram of a device according to an embodiment of the present invention.

[0034] Figure 8 FIG. 4 is a schematic structural block diagram of a modular system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0035] Exemplary embodiments of the present invention will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments may also be implemented in other forms and should not be construed as limiting the present invention. On the contrary, these embodiments are only used to fully and completely describe the present invention and fully convey the concept of the exemplary embodiments to those skilled in the art. The same reference numerals in the accompanying drawings represent the same or similar structures.

[0036] As mentioned above, there are currently a variety of needs for predicting or evaluating time-related parameters. Such parameters include, for example, one of the following: physiological parameters of the human body, physical parameters of a device or process, and chemical parameters of a device or process. The physiological parameters of the human body may include at least one of the following parameters: axial length, corneal curvature, lens diopter, diopter (degree of myopia), weight, and height. The common point of these parameters is that they change over time. However, the currently used technical means, such as the percentile curves of time-related parameters, have problems such as inaccuracy and failure to conform to practice well. In response to the above problems, the present invention proposes a feasible solution for achieving more accurate prediction and evaluation of time-related parameters.

[0037] In the present invention, firstly, parameter sample data as the basis for analysis is obtained. The parameter sample data may include a certain number of data groups (or may also be referred to as data pairs), each of which includes a parameter value of a parameter and a time value corresponding to the parameter value. In view of the above technical problems, the present invention proposes to consider the development and / or distribution law of the parameter itself, so that the mathematical model relied on for prediction or analysis can adapt to the characteristics of the parameter, thereby being more accurate and reliable. Here, the distribution pattern of the parameter at each time value may be considered, that is, the distribution pattern or distribution law of the parameter at any time value may be considered. The change pattern of the parameter over time may also be considered, that is, the way in which the parameter develops over time. Based on the acquired parameter sample data, a mathematical model for predicting and / or analyzing the parameter may be obtained in consideration of the above distribution pattern and change pattern. With this mathematical model, a more accurate prediction and / or evaluation result, especially one that conforms to the parameter characteristics, may be obtained.

[0038] Below, taking the relationship between the axial length of children's eyes and age as an example, a method for predicting or evaluating time-related parameters according to a preferred embodiment of the present invention is introduced. Of course, this method can also be used to predict or evaluate other time-related parameters. For example, parameters such as children's axial length, height, and weight will change in a growth-like manner as the body grows. Of course, there are also parameters that show random distribution characteristics over time. For example, the loss values ​​of some parts of some engineering machinery with very different operating conditions roughly meet random distribution over time.

[0039] like Figure 3As shown, the method for predicting or evaluating time-related parameters includes step S301, in which parameter sample data is obtained, and the parameter sample data includes a certain number of data groups, wherein each data group includes a parameter value of the parameter and a time value corresponding to the parameter value. Taking the axial length of a child as an example, each acquired data group includes the age and axial length of the child. The sample data can be obtained from a memory, for example, or read from other devices or input from a human-machine interface, etc.

[0040] The method also includes step S302, in which the distribution pattern of the parameter at each time value is determined or obtained. The distribution pattern of the parameter at each time value should be understood as the distribution pattern of the parameter at any time value. Taking the relationship between the axial length of children and age as an example, for children of the same age, their axial length has a randomly distributed distribution pattern. This distribution pattern can be, for example, a normal distribution, a binomial distribution, etc. "Determining" the distribution pattern should be understood as determining its distribution pattern by analyzing the parameter sample data. And "obtaining" the distribution pattern should be understood as obtaining the distribution pattern from other channels. For example, the distribution pattern may be set during the device initialization phase and stored in the memory, and the distribution pattern is read from the memory at this time. Alternatively, the distribution pattern may be input from the man-machine interface, or read from other devices. In the above and below, "determining" and "obtaining" should be understood similarly and will not be repeated.

[0041] The method also includes step S303, in which the temporal variation pattern of the parameter is determined or obtained. Taking the axial length of children as an example, the temporal variation pattern of the parameter should be understood as the variation pattern of the axial length of children as they grow older. For example, this variation pattern can be a medical growth pattern, such as a Gompertz function. Of course, there may also be other variation patterns, such as exponential functions, logarithmic functions, polynomial functions, S-type functions (sigmoid functions) or other growth curves. The variation pattern can be represented, for example, by the temporal variation of the characteristic quantity of the distribution pattern. The temporal variation of the characteristic quantity of the distribution pattern can be represented, for example, as one of the following functions of time: exponential function, logarithmic function, linear combination function, Gompertz function, S-type function.

[0042] The method further includes step S304, in which a constraint condition is determined or obtained. The constraint condition may include, for example, a value range of at least one of a parameter value and time in a probability function that depends on time. Of course, other constraints may also be included, such as stipulating that there is no conflict in the fitting results, that there are no singular points in the probability function, etc. In this embodiment, the constraint condition may include, for example, that the value range of the axial length is [22, 25] (mm), and the value range of time is [4, 18] (years).

[0043] The method further includes step S305, in which a probability function dependent on time is determined based on the parameter sample data, distribution mode, change mode and constraint conditions. Here, since the distribution characteristics of the parameter value at a specific time value and the change characteristics of the parameter over time are fully considered, and the actual application requirements are further adapted by considering the constraint conditions, the matching degree between the function used for fitting and the parameter itself is very high. Therefore, the obtained probability function can better reflect the actual distribution of the parameter.

[0044] In this application, "time-dependent probability function" should be understood as the following mapping relationship. Within the domain of the function, for any parameter value and any time, the corresponding probability density value (the probability of occurrence of the parameter value) or cumulative distribution value (the sum of the probabilities of occurrence of all values ​​less than or equal to the parameter value) of the parameter value at the time is given. In other words, the probability function can be either a probability density function or a cumulative distribution function. Both have their own advantages in predictive evaluation, and the function to be used can be selected as needed.

[0045] Here, determining the time-dependent probability function may include:

[0046] - based on the distribution pattern, generating a time-independent probability density function of the parameter, wherein the time-independent probability density function of the parameter comprises one or more characteristic quantities;

[0047] - generating a time-dependent probability density function by replacing each characteristic quantity of the time-independent probability density function of the parameter by a function of time;

[0048] - using the time-dependent probability density function as the time-dependent probability function; or integrating the time-dependent probability density function to obtain a time-dependent cumulative distribution function and using it as the time-dependent probability function.

[0049] Alternatively, a time-independent cumulative distribution function may be directly generated based on the distribution pattern, and the characteristic quantity of the time-independent cumulative distribution function may be replaced with a function of time to generate a time-dependent cumulative distribution function and use it as the time-dependent probability function.

[0050] In a preferred embodiment, the distribution mode is determined to be a normal distribution. As described above, for example, the distribution mode can be selected manually, or determined by the device analyzing the parameter sample data. In this example, the probability density function that is independent of time is expressed as:

[0051]

[0052] Where x is the parameter being analyzed, μ is the mean of the normal distribution, and σ is the standard deviation of the normal distribution. Here, the time-independent probability density function includes the characteristic quantities of the mean μ and the standard deviation σ. In particular, for physiological or pathological parameters of the human body, such as the axial length of children's eyes, the normal distribution has been verified to be a statistical function that conforms to reality.

[0053] As described above, the constraint condition may, for example, specify the range of values ​​of the parameter x. In this way, the trend of parameter changes can be better analyzed, and a larger range of specific parameter values ​​can be evaluated. Of course, the range of values ​​of the parameter x may also be specified according to specific circumstances. In this embodiment, the range of axial lengths that may occur in reality may be preferably selected for children, for example, [22, 25] (mm).

[0054] Next, the characteristic quantity of the probability density function that does not depend on time can be replaced by a function of time, where the mean μ can be replaced by a exp(b exp(ct)) and the standard deviation σ can be replaced by pt 2 +qt+r, where a, b, c, p, q and r are coefficients fitted based on parameter sample data. That is to say, here the mean μ is fitted based on the growth curve (Gompertz curve) and the standard deviation σ is fitted using a quadratic polynomial. Practice has shown that this fitting method is particularly suitable for a variety of human physiological or pathological parameters including children's axial length.

[0055] Therefore, the time-dependent probability density function can be expressed as:

[0056]

[0057] Here, the above probability density function can be directly used to predict the change trend of the parameter in the future and / or to evaluate the specific parameter value of the parameter. Of course, the time-dependent cumulative distribution function can also be obtained based on the probability density function:

[0058]

[0059] -Wherein, P(X,t)=Pr[x≤X] represents the probability that the parameter x takes a value less than or equal to X at time t.

[0060] In a preferred embodiment, a specific method of determining a time-dependent probability function may include:

[0061] - Determine the functional form of the time-dependent probability density function, such as Equation 2,

[0062] -Based on the parameter sample data, the probability density function is used for fitting analysis to determine the specific values ​​of the coefficients (a, b, c, p, q and r) of the probability density function, thereby obtaining a time-dependent probability density function.

[0063] - If necessary, obtain the time-dependent cumulative distribution function by integration.

[0064] Of course, it is also possible here to alternatively determine the time-dependent probability function in the following manner:

[0065] - Determine the functional form of the probability density function that is independent of time, such as Equation 1;

[0066] - determining the value of the characteristic quantity included in the probability density function independent of time for each time based on the parameter sample number;

[0067] - Based on the fitting analysis of each of the characteristic quantities (exp(b exp(ct)), pt 2 +qt+r) to obtain the time-dependent probability density function;

[0068] - If necessary, obtain the time-dependent cumulative distribution function by integration.

[0069] In the fitting analysis, for example, the least square method or the gradient descent method is used.

[0070] The time-dependent probability function can also be determined in other ways. For example, the time-dependent probability function can be directly determined according to the distribution pattern, change pattern and constraint conditions, and fitted based on the parameter sample data. For example, if the distribution pattern of the parameter is determined to be a normal distribution, and its change pattern over time is also determined to be a normal distribution. Here, the binary normal distribution function can be directly referenced, and the parameter sample data can be used for fitting analysis. For example, the part loss value of some engineering machinery components with very different operating conditions may satisfy the above distribution. Of course, those skilled in the art can conceive of other fitting or modeling methods based on the prior art.

[0071] The method further comprises step S306, in which the change trend of the parameter in the future is predicted and / or the specific parameter value of the parameter is evaluated. Here, for example, the refractive power trend of the individual child in the next year or years can be predicted based on the current axial length of the individual child. Of course, it is also possible to evaluate whether the current development is abnormal based on the current axial length of the individual child, for example, based on the difference between the current axial length of the individual child and the reference value.

[0072] In addition, considering that in some fields, people have adapted to or are accustomed to using percentile curves to predict or judge parameters. Here, a quasi-percentile curve graph can also be generated based on the time-dependent probability function, and prediction or evaluation can be performed based on the quasi-percentile curve graph. This quasi-percentile curve graph has a form similar to the percentile curve of the prior art, but because it is generated based on the probability function, it has higher accuracy and is more in line with the characteristics of the parameter itself.

[0073] Although the above steps are shown in a serial manner in the accompanying drawings, those skilled in the art will appreciate that these steps do not necessarily have to be performed in a serial manner. Some of the steps can be performed in parallel, for example, steps S302 to S304 can be performed in parallel. Based on the accompanying drawings and the methods shown, those skilled in the art can appreciate other implementations not shown or described herein without departing from the scope of the present invention.

[0074] Based on this embodiment, since the distribution pattern of the parameters and their temporal variation pattern are accurately set and the constraints are taken into account, the statistical method of the parameters is consistent with actual cognition, especially general cognition or research findings. Therefore, the prediction of any individual parameter can be close to the preset growth pattern and consistent with general cognition or research findings. At the same time, the extrapolation is improved and the stability is increased, and the natural smoothness of the obtained probability function will not have obvious errors of intersections of different percentiles. In addition, the compatibility with extreme values ​​is improved, and any extreme value of the parameter sample data can be included in the probability function, so that the probability function can be applied to any patient even without any extrapolation.

[0075] exist Figures 4 to 6 The probability density functions of the axial length obtained based on the above-mentioned embodiments are shown in FIG. Figure 4 ), cumulative distribution function ( Figure 5 ), and the quasi-percentile curves ( Figure 6 ). Figure 4 An example of the time-dependent probability density function (Formula 2) obtained according to the above method is shown. Figure 4 As shown in FIG. 1 , the probability density function obtained by this method is a continuous function, which exhibits a continuous surface. Based on this surface, the probability value of any parameter value within the domain of the function at any time point can be determined.

[0076] Figure 5 An example of a time-dependent cumulative distribution function is shown. The graphical representation of the cumulative distribution function also shows a continuous surface. Based on the surface, for any parameter value within the domain of the function, the total probability of all parameters being less than or equal to the parameter value at any point in time can be determined.

[0077] In addition, Figure 5 The cumulative distribution function surface shown in the figure shows a number of percentile-like curves. The following uses P25 as an example to introduce how to generate these percentile-like curves. First, determine a plane that is perpendicular to the cumulative density axis (the vertical axis in the figure) and intersects the axis at the 25% scale (see Figure 5 ). The intersection of the plane and the cumulative distribution function surface forms the P25 curve in the percentile-like curve. The generation methods of other percentile-like curves are similar and will not be repeated here. Figure 6 Each percentile curve is shown separately in .

[0078] like Figure 4 and Figure 5 As shown in the figure, due to the appropriate selection of distribution mode and change mode, the probability density function curve and the cumulative distribution function curve surface are consistent with the general consensus of children's axial length. Moreover, since the definition domain of this function is large, it shows a good extrapolation effect outside the sample data. Figure 6 The percentile curve shown here avoids the existing technology (see Figure 1-2 ) in the problem Figure 6 There is no longer a situation where different percentile lines intersect. In this method embodiment, due to the inclusiveness of the mathematical model used for fitting, that is, the probability function itself, any value in the original sample data can be included in the fitting model. Therefore, the model is applicable to any patient. And because the domain of the probability function itself (the range of parameter values ​​and time) can be selected to be larger, there is no need for any extrapolation. Moreover, because the distribution pattern, change pattern, and constraint conditions take into account the characteristics of the parameters themselves, the establishment of the probability function is less dependent on the amount of sample data. In other words, even if the amount of sample data is small, it is possible to establish a probability function that meets the characteristics of the parameter.

[0079] In an exemplary embodiment of the present invention, a computer-readable storage medium is further provided, on which a computer program is stored, the program including executable instructions, which can implement the steps of the method described in any of the above embodiments when the executable instructions are executed by, for example, a processor. In some possible implementations, various aspects of the present invention can also be implemented in the form of a program product, which includes a program code, and when the program product is run on a terminal device, the program code is used to enable the terminal device to execute the steps of various exemplary embodiments described above for the method.

[0080] The program product for realizing the above method according to an embodiment of the present invention can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can be, for example, but not limited to, a system, device or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0081] The program code for performing the operations of the present application may be written in any combination of one or more programming languages ​​and may be executed entirely on the user's computing device, partially on the user's computing device, as a stand-alone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0082] In an exemplary embodiment of the present application, a device for predicting or evaluating a time-related parameter is also provided, and the device may include a processor and a memory for storing executable instructions of the processor. The processor is configured to execute the steps of the method for predicting a time-related parameter in any of the above embodiments by executing the executable instructions.

[0083] Refer to the following Figure 7 800 according to this embodiment of the present application is described. Figure 7 The device 700 shown is merely an example and should not limit the functions and scope of use of the embodiments of the present application.

[0084] like Figure 7 As shown, the device 700 is in the form of a general-purpose computing device. The components of the device 700 may include, but are not limited to: at least one processing unit 710, at least one storage unit 720, a bus 730 connecting different system components (including the storage unit 720 and the processing unit 710), a display unit 740, etc.

[0085] The storage unit 720 stores program codes, which can be executed by the processing unit 710, so that the processing unit 710 executes steps S301-S306 described above for the method embodiment.

[0086] The storage unit 720 may include a readable medium in the form of a volatile storage unit, such as a random access storage unit (RAM) 7201 and / or a cache storage unit 7202 , and may further include a read-only storage unit (ROM) 7203 .

[0087] The storage unit 720 may also include a program / utility 7204 having a set (at least one) of program modules 7205, such program modules 7205 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0088] The device 700 may also communicate with one or more external devices 1000 (e.g., keyboards, pointing devices, Bluetooth devices, etc.), one or more devices that enable a user to interact with the device 700, and / or any device that enables the device 700 to communicate with one or more other computing devices (e.g., routers, modems, etc.). Such communication may be performed via an input / output (I / O) interface 750. The device 700 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) via a network adapter 760. The network adapter 760 may communicate with other modules of the device 700 via a bus 730. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the device 700, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0089] Through the description of the above implementations, it is easy for those skilled in the art to understand that the example implementations described here can be implemented by software (as described above) or by combining software with necessary hardware.

[0090] like Figure 8As shown, the present application also relates to a modular system 800 for predicting or evaluating time-related parameters, which includes a data acquisition unit 810 configured to obtain parameter sample data. The data acquisition unit can be constructed as an I / O interface, a human-computer interaction interface, a software module with a reading function, a data acquisition device, etc. The modular system 800 also includes a data analysis unit 820, which is configured to determine or obtain a distribution pattern of the parameter at each time value, determine or obtain a change pattern of the parameter over time, determine or obtain a constraint condition, and determine a probability function that depends on time based on the parameter sample data, the distribution pattern, the change pattern, and the constraint condition. The data analysis unit 820 can include, for example, a storage device and a processing device, but can also include an I / O interface, etc. The modular system can also include a prediction or evaluation unit 830, which is configured to predict the change trend of the parameter in the future time based on the time-dependent probability function and / or evaluate the specific parameter value of the parameter. The prediction or evaluation unit 830 can be constructed as a display device or an output device, etc.

[0091] The features or advantages described above based on the method are also applicable to the storage medium, device and modular system according to the present invention, and are therefore not described in detail.

[0092] Those skilled in the art will appreciate that various aspects of the present application may be implemented as a system, method or program product. Therefore, various aspects of the present application may be specifically implemented in the following forms, namely: a complete hardware implementation, a complete software implementation (including firmware, microcode, etc.), or a combination of hardware and software, which may be collectively referred to as "circuit", "module" or "system" herein.

[0093] Those skilled in the art will readily appreciate other embodiments of the present application after considering the description and practicing the contents disclosed herein. The present application is intended to cover any modification, use or adaptation of the present application, which follows the general principles of the present application and includes common knowledge or customary techniques in the art that are not disclosed in the present application. The description and embodiments are intended to be exemplary only, and the true scope and spirit of the present application are indicated by the appended claims.

Claims

1. A device for predicting or estimating a time-dependent parameter, characterized in that The device comprises: Processor; and A memory, configured to store executable instructions of the processor; The processor is configured to execute the executable instructions to implement the following steps: - obtaining parameter sample data, wherein the parameter sample data includes a certain number of data groups, each data group includes a parameter value of a parameter and a time value corresponding to the parameter value; - determining or obtaining the distribution pattern of the parameter at each time value; - determining or obtaining a temporal variation pattern of said parameter; -determine or obtain constraints; -determining a time-dependent probability function based on the parameter sample data, the distribution pattern, the variation pattern, and the constraint condition, wherein the constraint condition includes a value range of at least one of a parameter value and time of the parameter in the time-dependent probability function; and - predicting the variation trend of the parameter at a future time and / or evaluating a specific parameter value of the parameter based on the time-dependent probability function.

2. The device according to claim 1, characterized in that The variation pattern is represented by the variation of the characteristic quantity of the distribution pattern over time; The temporal variation of the characteristic quantity of the distribution pattern is represented by one of the following functions of time: an exponential function, a logarithmic function, a linear combination function, a Gompertz function, and a S-type function.

3. The device according to claim 2, characterized in that Determining the time-dependent probability function involves: - based on the distribution pattern, generating a time-independent probability density function of the parameter, wherein the time-independent probability density function of the parameter comprises one or more characteristic quantities; - replacing each of the one or more feature quantities with a function of time to generate a probability density function that depends on time; - using the time-dependent probability density function as the time-dependent probability function; or integrating the time-dependent probability density function to obtain a time-dependent cumulative distribution function and using it as the time-dependent probability function.

4. The device according to claim 2, characterized in that Determining the time-dependent probability function involves: - generating a time-independent cumulative distribution function of the parameter based on the distribution pattern, wherein the time-independent cumulative distribution function of the parameter includes one or more characteristic quantities; - By replacing each of the one or more feature quantities with a function of time, a cumulative distribution function dependent on time is generated and used as the probability function dependent on time.

5. The device according to claim 3, characterized in that Generating a time-independent probability density function of the parameter based on the distribution pattern further comprises: - The distribution pattern is determined or obtained as a normal distribution, and the probability density function that is independent of time is expressed as: -where x is the parameter, μ is the mean, σ is the standard deviation, -The characteristic quantities are mean μ and standard deviation σ.

6. The device according to claim 5, characterized in that Replacing the characteristic quantity of the probability density function that is independent of time with a function of time includes: -Replace the mean μ with a exp(b exp(ct)), -Replace the standard deviation σ with pt 2 +qt+r, Where t is time, a, b, c, p, q and r are coefficients generated based on parameter sample data fitting. -The time-dependent probability density function is expressed as:

7. The device according to claim 6, characterized in that The time-dependent cumulative distribution function is expressed as: -Where P(X,t)=Pr[x≤X,t] represents the probability that the parameter x takes a value less than or equal to X at time t.

8. The device according to claim 1, characterized in that Determining a time-dependent probability function based on the parameter sample data, the distribution pattern, the variation pattern, and the constraint condition includes: - performing a fitting analysis using said parameter sample data to determine said time-dependent probability function.

9. The device according to claim 8, characterized in that Fitting analysis was performed using the least squares method or gradient descent method.

10. The device according to claim 1, characterized in that Predicting the change trend of the parameter in the future based on the time-dependent probability function and / or evaluating the specific parameter value of the parameter further includes: generating a class percentile graph based on the time-dependent probability function, and The variation trend is predicted based on the class percentile graph and / or a specific parameter value of the parameter is evaluated.

11. The device according to claim 1, characterized in that The parameter includes one of the following: a physiological parameter of a human body, a physical parameter of a device or process, a chemical parameter of a device or process; The physiological parameters of the human body include at least one of the following parameters: axial length, corneal curvature, lens diopter, diopter, weight, and height.

12. A method for predicting or estimating a time-dependent parameter, comprising: - obtaining parameter sample data, wherein the parameter sample data includes a certain number of data groups, each data group includes a parameter value of a parameter and a time value corresponding to the parameter value; - determining or obtaining the distribution pattern of the parameter at each time value; - determining or obtaining a temporal variation pattern of said parameter; -determine or obtain constraints; -determining a time-dependent probability function based on the parameter sample data, the distribution pattern, the variation pattern, and the constraint condition, wherein the constraint condition includes a value range of at least one of a parameter value and time of the parameter in the time-dependent probability function; and - predicting the variation trend of the parameter at a future time and / or evaluating a specific parameter value of the parameter based on the time-dependent probability function.

13. The method according to claim 12, characterized in that The variation pattern is represented by the variation of the characteristic quantity of the distribution pattern over time; The temporal variation of the characteristic quantity of the distribution pattern is represented by one of the following functions of time: an exponential function, a logarithmic function, a linear combination function, a Gompertz function, and a S-type function.

14. The method according to claim 13, characterized in that Determining the time-dependent probability function involves: - based on the distribution pattern, generating a time-independent probability density function of the parameter, wherein the time-independent probability density function of the parameter comprises one or more characteristic quantities; - replacing each of the one or more feature quantities with a function of time to generate a probability density function that depends on time; - using the time-dependent probability density function as the time-dependent probability function; or integrating the time-dependent probability density function to obtain a time-dependent cumulative distribution function and using it as the time-dependent probability function.

15. The method according to claim 13, characterized in that Determining the time-dependent probability function involves: - generating a time-independent cumulative distribution function of the parameter based on the distribution pattern, wherein the time-independent cumulative distribution function of the parameter includes one or more characteristic quantities; - By replacing each of the one or more feature quantities with a function of time, a cumulative distribution function dependent on time is generated and used as the probability function dependent on time.

16. The method according to claim 14, characterized in that Generating a time-independent probability density function of the parameter based on the distribution pattern further comprises: - The distribution pattern is determined or obtained as a normal distribution, and the probability density function that is independent of time is expressed as: -where x is the parameter, μ is the mean, σ is the standard deviation, -The characteristic quantities are mean μ and standard deviation σ.

17. The method according to claim 16, characterized in that Replacing the characteristic quantity of the probability density function that is independent of time with a function of time includes: -Replace the mean μ with a exp(b exp(ct)); -Replace the standard deviation σ with pt 2 +qt+r, Where t is time, a, b, c, p, q and r are coefficients generated based on parameter sample data fitting. -The time-dependent probability density function is expressed as:

18. The method according to claim 17, characterized in that The time-dependent cumulative distribution function is expressed as: -Wherein, P(X,T)=Pr[x≤X] represents the probability that the parameter x takes a value less than or equal to X at time t.

19. The method according to claim 12, characterized in that Determining a time-dependent probability function based on the parameter sample data, the distribution pattern, the variation pattern, and the constraint condition includes: - performing a fitting analysis using said parameter sample data to determine said time-dependent probability function.

20. The method according to claim 19, characterized in that Fitting analysis was performed using the least squares method or gradient descent method.

21. The method according to claim 12, characterized in that Predicting the change trend of the parameter in the future based on the time-dependent probability function and / or evaluating the specific parameter value of the parameter further includes: generating a class percentile graph based on the time-dependent probability function, and The variation trend is predicted based on the class percentile graph and / or a specific parameter value of the parameter is evaluated.

22. The method according to claim 12, characterized in that The parameter includes one of the following: a physiological parameter of a human body, a physical parameter of a device or process, a chemical parameter of a device or process; The physiological parameters of the human body include at least one of the following parameters: axial length, corneal curvature, lens diopter, diopter, weight, and height.

23. A computer-readable storage medium having a computer program stored thereon, the computer program comprising executable instructions, and when the executable instructions are executed by a processor, the method according to any one of claims 12 to 22 is implemented.

Images