Intraocular lens power prediction method, device and storage medium
By combining a multilayer perceptron regression model with the SRK/T formula and specific coding processing, the problem of insufficient axial length and IOL model compatibility in intraocular lens refractive prediction was solved, achieving high-precision and stable refractive prediction results.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- ZHONGSHAN OPHTHALMIC CENT SUN YAT SEN UNIV
- Filing Date
- 2026-06-03
- Publication Date
- 2026-07-14
AI Technical Summary
Existing technologies for intraocular lens (IOL) refractive prediction are not well adapted, especially for different axial lengths and IOL models, resulting in unstable prediction accuracy and limited precision.
A multilayer perceptron regression model combined with the SRK/T formula was adopted. By collecting patients' biological eye measurement parameters and intraocular lens constants, a sample set was constructed and trained to obtain spherical equivalent residual data. The multilayer perceptron regression model was used for prediction, and sine-cosine periodic coding and one-hot coding were combined to improve the model's learning ability and cross-model prediction ability.
It significantly improves the accuracy of intraocular lens refractive prediction, especially in patients with long axial lengths where the mean absolute error is as low as 0.24D, and enhances the accuracy and stability of cross-type prediction.
Smart Images
Figure CN122393005A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of data prediction technology for intraocular lenses, specifically to a method, device, and storage medium for predicting refractive power in intraocular lenses. Background Technology
[0002] Cataracts are one of the leading causes of reversible vision impairment worldwide, and surgery remains the only effective treatment. With advancements in surgical techniques and optical biometry equipment, cataract surgery has gradually shifted from traditional vision restoration to refractive surgery, making postoperative refractive accuracy a crucial indicator of surgical quality. Patients' increasing demands for postoperative uncorrected visual acuity have made the accuracy and stability of intraocular lens (IOL) power calculations increasingly important, driving the continuous development of related calculation formulas and predictive models.
[0003] Currently, widely used third-generation IOL calculation formulas in clinical practice include SRK / T, HofferQ, Holladay1, and Haigis. These formulas are based on thin lens theory and statistical regression methods, with their core function being the prediction of the effective lens position (ELP) of the intraocular lens. In recent years, the Barrett Universal II (BUII) formula has introduced a more refined optical model and theoretical algorithm, gaining widespread clinical acceptance. Furthermore, the EVO formula incorporates thick lens optics and posterior corneal astigmatism prediction into its calculation framework, further improving the accuracy of refractive prediction.
[0004] However, current refractive prediction still has the following shortcomings: the prediction accuracy is unstable for different axial lengths (especially short and long axial lengths); the compatibility between different IOL models is insufficient, the model prediction performance is reduced, resulting in limited prediction accuracy. Summary of the Invention
[0005] The purpose of this application is to overcome the shortcomings and deficiencies in the prior art and provide a method, device and storage medium for predicting the refractive error of an intraocular lens, which can accurately predict the refractive error of an intraocular lens.
[0006] The first aspect of this application provides a method for predicting refractive error in an intraocular lens, including:
[0007] We collected bio-eye measurement parameters, intraocular lens constant, intraocular lens refractive power, and actual spherical equivalent data after cataract surgery from multiple patients.
[0008] The baseline predicted spherical equivalent data for each patient is calculated based on the SRK / T formula, and the spherical equivalent residual data is obtained based on the actual spherical equivalent data and the baseline predicted spherical equivalent data.
[0009] A sample set was constructed based on the bio-eye measurement parameters, intraocular lens constant, intraocular lens refractive power, and spherical lens equivalent residual data of the aforementioned multiple patients;
[0010] Using the biological eye measurement parameters, intraocular lens constant, and intraocular lens refractive power of the sample set as input features, and the spherical lens equivalent residual data as output features, the multilayer perceptron regression model is trained to obtain the spherical lens equivalent residual prediction model.
[0011] The target biological eye measurement parameters, target intraocular lens constant, and target intraocular lens refractive power of the target patient are input into the spherical lens equivalent residual prediction model to obtain the spherical lens equivalent prediction residual value.
[0012] Based on the baseline predicted spherical equivalent data and the spherical equivalent prediction residual value of the target patient, the intraocular lens refractive prediction result of the target patient is obtained.
[0013] In one implementation, the biological eye measurement parameters include axial length, corneal curvature, anterior chamber depth, corneal thickness, lens thickness, and corneal diameter.
[0014] As one implementation method, after using the biological eye measurement parameters, intraocular lens constant, and intraocular lens refractive power of the sample set as input features, the method further includes the following steps:
[0015] The input features are Z-score standardized, the intraocular lens constant is sine-cosine periodic encoded, and the eye is uniquely encoded; the eye is divided into left eye and right eye.
[0016] As one implementation method, the steps of training a multilayer perceptron regression model to obtain a spherical mirror equivalent residual prediction model include:
[0017] The sample set is divided into a training set, a validation set, and a test set; wherein the number of samples in the training set is greater than that in the validation set, and the number of samples in the validation set is greater than or equal to that in the test set.
[0018] The multilayer perceptron regression model is trained based on the training set to obtain the first residual prediction model;
[0019] Based on the validation set, the parameters of the first residual prediction model are tuned to obtain the second residual prediction model;
[0020] The second residual prediction model is tested using the test set to obtain a spherical mirror equivalent residual prediction model that passes the test.
[0021] As one implementation, the multilayer perceptron regression model includes two or more fully connected hidden layers, each layer employing the swish activation function, and randomly dropping neuron outputs at a rate of 20%–40% in the hidden layers to achieve dropout regularization.
[0022] As one implementation method, the step of obtaining spherical equivalent residual data based on the actual spherical equivalent data and the basic predicted spherical equivalent data includes:
[0023] The equivalent residual data of the spherical mirror can be obtained using the following formula:
[0024] ; in, For spherical mirror equivalent residual data, This is the actual spherical equivalent data after surgery. This is the basis for predicting the equivalent spherical mirror data of the sample set.
[0025] As one implementation method, the step of obtaining the intraocular lens refractive prediction result for the target patient based on the target patient's baseline predicted spherical equivalent data and the spherical equivalent prediction residual value includes:
[0026] The refractive prediction result of the intraocular lens can be obtained using the following formula:
[0027] ; in, This is the result of intraocular lens refractive prediction. The baseline predictive spherical equivalent data for the target patients, The residual value is the equivalent prediction value of the spherical mirror.
[0028] Compared to related technologies, the intraocular lens (IOL) refractive error prediction method of this application collects bio-eye measurement parameters, IOL constants, IOL refractive power, and actual spherical equivalent data after cataract surgery from multiple patients. Based on the SRK / T formula, it calculates the baseline predicted spherical equivalent data for each patient. Using the actual spherical equivalent data and the baseline predicted spherical equivalent data, it obtains spherical equivalent residual data for nonlinear structures. This residual data is then used to train a multilayer perceptron regression model, resulting in a spherical equivalent residual prediction model specifically designed to predict spherical equivalent residual data for nonlinear structures. The predicted spherical equivalent residual value for the target patient, calculated using the spherical equivalent residual prediction model, along with the baseline predicted spherical equivalent data for the target patient based on the SRK / T formula, yields the IOL refractive error prediction result for the target patient. Because it combines the spherical equivalent prediction residual value obtained from nonlinear dimension prediction with the baseline predicted spherical equivalent data calculated from the linear dimension, it can accurately predict IOL refractive errors.
[0029] A second aspect of this application provides an intraocular lens refractive prediction device, comprising:
[0030] The data acquisition module is used to collect bio-eye measurement parameters, intraocular lens constant, intraocular lens refractive power, and actual spherical equivalent data after cataract surgery from multiple patients.
[0031] The residual data acquisition module calculates the baseline predicted spherical equivalent data for each patient based on the SRK / T formula, and acquires the spherical equivalent residual data based on the actual spherical equivalent data and the baseline predicted spherical equivalent data.
[0032] The sample set construction module is used to construct a sample set based on the bio-eye measurement parameters, intraocular lens constant, intraocular lens refractive power, and spherical equivalent residual data of the multiple patients;
[0033] The residual prediction model acquisition module is used to train the multilayer perceptron regression model by taking the biological eye measurement parameters, intraocular lens constant, and intraocular lens refractive power of the sample set as input features and the spherical equivalent residual data as output features, to obtain the spherical equivalent residual prediction model.
[0034] The residual prediction module is used to input the target biological eye measurement parameters, target intraocular lens constant, and target intraocular lens refractive power of the target patient into the spherical equivalent residual prediction model to obtain the spherical equivalent prediction residual value.
[0035] The refractive prediction result acquisition module is used to obtain the intraocular lens refractive prediction result of the target patient based on the target patient's baseline predicted spherical equivalent data and the spherical equivalent prediction residual value.
[0036] In one implementation, the biological eye measurement parameters include axial length, corneal curvature, anterior chamber depth, corneal thickness, lens thickness, and corneal diameter.
[0037] A third aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the intraocular lens refractive prediction method as described above.
[0038] To provide a clearer understanding of this application, the specific embodiments of this application will be described below in conjunction with the accompanying drawings. Attached Figure Description
[0040] Figure 1 This is a flowchart of an intraocular lens refractive prediction method according to an embodiment of this application.
[0041] Figure 2 This is a schematic diagram of the training process of a multilayer perceptron regression model according to an embodiment of this application.
[0042] Figure 3 This is a schematic diagram of the module connection of an intraocular lens refractive prediction device according to an embodiment of this application.
[0043] 100. Intraocular lens refractive prediction device; 101. Data acquisition module; 102. Residual data acquisition module; 103. Sample set construction module; 104. Residual prediction model acquisition module; 105. Residual prediction module; 106. Refractive prediction result acquisition module. Detailed Implementation
[0045] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.
[0046] It should be understood that the described embodiments are merely some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of the embodiments of this application.
[0047] In the following description, when referring to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. In the description of this application, it should be understood that the terms "first," "second," "third," etc., are used only to distinguish similar objects and are not necessarily used to describe a specific order or sequence, nor should they be construed as indicating or implying relative importance. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances. The singular forms "a," "the," and "the" used in this application and the appended claims are also intended to include the plural forms, unless the context clearly indicates otherwise. The word "if" as used herein can be interpreted as "when," "when," or "in response to determination."
[0048] Furthermore, in the description of this application, unless otherwise stated, "multiple" means two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. The character " / " generally indicates that the preceding and following related objects have an "or" relationship.
[0049] Please see Figure 1 This is a flowchart of the intraocular lens refractive prediction method according to the first embodiment of this application. The method includes:
[0050] S1: Collect bio-eye measurement parameters, intraocular lens constant, intraocular lens refractive power, and actual spherical equivalent data after cataract surgery from multiple patients.
[0051] The biological eye measurement parameters include, but are not limited to, axial length (AL), corneal curvature (K1 / K2), anterior chamber depth (ACD), lens thickness (LT), central corneal thickness (CCT), corneal diameter (WTW), intraocular lens power (IOLpower), and intraocular lens constant (Aconstant).
[0052] S2: Calculate the baseline predicted spherical equivalent data (SRK / T predicted SE) for each patient based on the SRK / T formula, and obtain the spherical equivalent residual data based on the actual spherical equivalent data and the baseline predicted spherical equivalent data.
[0053] The SRK / T formula is a traditional optical formula used to calculate the power of the intraocular lens (IOL) during cataract surgery and is widely used in clinical practice. It combines a theoretical eye model with regression analysis to predict postoperative refractive status relatively accurately.
[0054] The equivalent residual data of the spherical mirror can be obtained using the following formula:
[0055] ;
[0056] in, For spherical mirror equivalent residual data, This represents the actual spherical equivalent data (actual SE) after surgery. This is the basis for predicting the equivalent spherical mirror data of the sample set.
[0057] S3: Construct a sample set based on the bio-eye measurement parameters, intraocular lens constant, intraocular lens refractive power, and the spherical equivalent residual data of the multiple patients.
[0058] S4: Using the biological eye measurement parameters, intraocular lens constant, and intraocular lens refractive power of the sample set as input features, and the spherical lens equivalent residual data as output features, the multilayer perceptron regression model (MLP model) is trained to obtain the spherical lens equivalent residual prediction model.
[0059] The multilayer perceptron regression model contains two or more fully connected hidden layers. Each layer uses the swish activation function, and 20%–40% of the neuron outputs are randomly dropped in the hidden layers to achieve dropout regularization. The model is trained using the SGD optimizer.
[0060] After training, cross-validation can be used to verify the reliability of the model, ensuring stable predictive performance across different axial length ranges (e.g., 21–35 mm).
[0061] S5: Input the target biological eye measurement parameters, target intraocular lens constant, and target intraocular lens refractive power of the target patient into the spherical lens equivalent residual prediction model to obtain the spherical lens equivalent prediction residual value.
[0062] S6: Based on the target patient's baseline predicted spherical equivalent data and the spherical equivalent prediction residual value, obtain the target patient's intraocular lens refractive prediction result.
[0063] The refractive prediction result of the intraocular lens can be obtained using the following formula:
[0064] ;
[0065] in, This is the result of intraocular lens refractive prediction. The baseline predictive spherical equivalent data for the target patients, The residual value is the equivalent prediction value of the spherical mirror.
[0066] The core of this invention lies in the following: First, the basic predicted spherical equivalent (SE) is calculated using the SRK / T formula. Then, an MLP model is used to predict the residual, which is the difference between the actual SE and the SRK / T predicted SE. The residual learning strategy avoids the pressure of directly fitting the physical optical laws to the deep learning model, allowing the model to focus on learning nonlinear relationships that the formula cannot characterize, thereby significantly improving prediction accuracy.
[0067] In a feasible embodiment, after using the biological eye measurement parameters, intraocular lens constant, and intraocular lens refractive power of the sample set as input features, the method further includes the following steps:
[0068] The input features are Z-score standardized, the intraocular lens constant is sine-cosine periodic encoded, and the eye is one-hot encoded; the eye is divided into left eye and right eye.
[0069] The aforementioned encoding method enhances the model's learning ability across different parameter scales and periodic structures, preventing the model from getting trapped in local optima. Furthermore, encoding the IOL constant using a sine-cosine periodic embedding method enables the deep learning model to identify periodic correlations between different IOL types, improving cross-type prediction capabilities. Additionally, by incorporating optical structural parameters such as ELP, IOL diopter, and corneal diameter, the model acquires sufficient physical and optical background information.
[0070] Please see Figure 2In one feasible embodiment, the step of training a multilayer perceptron regression model to obtain a spherical mirror equivalent residual prediction model includes:
[0071] S41: Divide the sample set into a training set, a validation set, and a test set; wherein the number of samples in the training set is greater than that in the validation set, and the number of samples in the validation set is greater than or equal to that in the test set.
[0072] During training, GPU-accelerated training can be used, and mini-batch stochastic gradient descent algorithm can be employed for optimization to improve training efficiency.
[0073] S42: Train the multilayer perceptron regression model based on the training set to obtain the first residual prediction model.
[0074] S43: Based on the validation set, the parameters of the first residual prediction model are tuned to obtain the second residual prediction model.
[0075] S44: Test the second residual prediction model using the test set to obtain a spherical mirror equivalent residual prediction model that passes the test.
[0076] For example, through training and validation in a large sample cohort covering the entire axial length range of 21–35 mm, the method of this invention demonstrated significantly better predictive ability than conventional formulas and network models for direct intraocular lens refractive prediction in a large-scale test set encompassing the entire axial length range of 21–35 mm. In the subgroup of patients with long axial lengths, the prediction method of this invention maintained stable performance, with a mean absolute error as low as 0.24 D.
[0077] Compared to related technologies, the intraocular lens (IOL) refractive error prediction method of this application collects bio-eye measurement parameters, IOL constants, IOL refractive power, and actual spherical equivalent data after cataract surgery from multiple patients. Based on the SRK / T formula, it calculates the baseline predicted spherical equivalent data for each patient. Using the actual spherical equivalent data and the baseline predicted spherical equivalent data, it obtains spherical equivalent residual data for nonlinear structures. This residual data is then used to train a multilayer perceptron regression model, resulting in a spherical equivalent residual prediction model specifically designed to predict spherical equivalent residual data for nonlinear structures. The predicted spherical equivalent residual value for the target patient, calculated using the spherical equivalent residual prediction model, along with the baseline predicted spherical equivalent data for the target patient based on the SRK / T formula, yields the IOL refractive error prediction result for the target patient. Because it combines the spherical equivalent prediction residual value obtained from nonlinear dimension prediction with the baseline predicted spherical equivalent data calculated from the linear dimension, it can accurately predict IOL refractive errors.
[0078] Please see Figure 3A second embodiment of this application provides an intraocular lens refractive prediction device, comprising:
[0079] The data acquisition module is used to collect bio-eye measurement parameters, intraocular lens constant, intraocular lens refractive power, and actual spherical equivalent data after cataract surgery from multiple patients; wherein, the bio-eye measurement parameters include axial length, corneal curvature, anterior chamber depth, corneal thickness, lens thickness, and corneal diameter.
[0080] The residual data acquisition module calculates the baseline predicted spherical equivalent data for each patient based on the SRK / T formula, and acquires the spherical equivalent residual data based on the actual spherical equivalent data and the baseline predicted spherical equivalent data.
[0081] The sample set construction module is used to construct a sample set based on the bio-eye measurement parameters, intraocular lens constant, intraocular lens refractive power, and spherical equivalent residual data of the multiple patients.
[0082] The residual prediction model acquisition module is used to train the multilayer perceptron regression model by taking the biological eye measurement parameters, intraocular lens constant, and intraocular lens refractive power of the sample set as input features and the spherical equivalent residual data as output features, to obtain the spherical equivalent residual prediction model.
[0083] The residual prediction module is used to input the target biological eye measurement parameters, target intraocular lens constant, and target intraocular lens refractive power of the target patient into the spherical equivalent residual prediction model to obtain the spherical equivalent prediction residual value.
[0084] The refractive prediction result acquisition module is used to obtain the intraocular lens refractive prediction result of the target patient based on the target patient's baseline predicted spherical equivalent data and the spherical equivalent prediction residual value.
[0085] It should be noted that the intraocular lens refractive prediction device provided in the second embodiment of this application is only illustrated by the above-described division of functional modules when performing the intraocular lens refractive prediction method. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the intraocular lens refractive prediction device provided in the second embodiment of this application and the intraocular lens refractive prediction method of the first embodiment of this application belong to the same concept, and its implementation process is detailed in the method embodiment, which will not be repeated here.
[0086] A third aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the intraocular lens refractive prediction method as described above.
[0087] The device embodiments described above are merely illustrative. The components described as separate parts 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 the modules can be selected to achieve the purpose of this application according to actual needs. Those skilled in the art can understand and implement this without any inventive effort.
[0088] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0089] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 The computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function selected in one or more boxes.
[0090] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function selected in one or more boxes.
[0091] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.
[0092] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.
[0093] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.
[0094] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.
[0095] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.
Claims
1. A method for predicting refractive error in an intraocular lens, characterized in that, include: We collected bio-eye measurement parameters, intraocular lens constant, intraocular lens refractive power, and actual spherical equivalent data after cataract surgery from multiple patients. The baseline predicted spherical equivalent data for each patient is calculated based on the SRK / T formula, and the spherical equivalent residual data is obtained based on the actual spherical equivalent data and the baseline predicted spherical equivalent data. A sample set was constructed based on the bio-eye measurement parameters, intraocular lens constant, intraocular lens refractive power, and spherical lens equivalent residual data of the aforementioned multiple patients; Using the biological eye measurement parameters, intraocular lens constant, and intraocular lens refractive power of the sample set as input features, and the spherical lens equivalent residual data as output features, the multilayer perceptron regression model is trained to obtain the spherical lens equivalent residual prediction model. The target biological eye measurement parameters, target intraocular lens constant, and target intraocular lens refractive power of the target patient are input into the spherical lens equivalent residual prediction model to obtain the spherical lens equivalent prediction residual value. Based on the baseline predicted spherical equivalent data and the spherical equivalent prediction residual value of the target patient, the intraocular lens refractive prediction result of the target patient is obtained.
2. The method for predicting refractive error of an intraocular lens according to claim 1, characterized in that, The biological eye measurement parameters include axial length, corneal curvature, anterior chamber depth, corneal thickness, lens thickness, and corneal diameter.
3. The method for predicting refractive error of an intraocular lens according to claim 1, characterized in that, After using the biological eye measurement parameters, intraocular lens constant, and intraocular lens refractive power of the sample set as input features, the following steps are also included: The input features are Z-score standardized, the intraocular lens constant is sine-cosine periodic encoded, and the eye is uniquely encoded; the eye is divided into left eye and right eye.
4. The method for predicting intraocular lens refractive error according to claim 1, characterized in that, The steps for training a multilayer perceptron regression model to obtain a spherical mirror equivalent residual prediction model include: The sample set is divided into a training set, a validation set, and a test set; wherein the number of samples in the training set is greater than that in the validation set, and the number of samples in the validation set is greater than or equal to that in the test set. The multilayer perceptron regression model is trained based on the training set to obtain the first residual prediction model; Based on the validation set, the parameters of the first residual prediction model are tuned to obtain the second residual prediction model; The second residual prediction model is tested using the test set to obtain a spherical mirror equivalent residual prediction model that passes the test.
5. The method for predicting refractive error of an intraocular lens according to claim 1, characterized in that, The multilayer perceptron regression model contains two or more fully connected hidden layers. Each layer uses the swish activation function, and 20%–40% of the neuron outputs are randomly dropped in the hidden layers to achieve dropout regularization.
6. The method for predicting refractive error of an intraocular lens according to claim 1, characterized in that, The step of obtaining spherical equivalent residual data based on the actual spherical equivalent data and the basic predicted spherical equivalent data includes: The equivalent residual data of the spherical mirror can be obtained using the following formula: ； in, For spherical mirror equivalent residual data, This is the actual spherical equivalent data after surgery. This is the basis for predicting the equivalent spherical mirror data of the sample set.
7. The method for predicting refractive error of an intraocular lens according to claim 1, characterized in that, The steps for obtaining the intraocular lens refractive prediction result for the target patient based on the baseline predicted spherical equivalent data and the spherical equivalent prediction residual value include: The refractive prediction result of the intraocular lens can be obtained using the following formula: ； in, This is the result of intraocular lens refractive prediction. The baseline predictive spherical equivalent data for the target patients, The residual value is the equivalent prediction value of the spherical mirror.
8. A refractive prediction device for an intraocular lens, characterized in that, include: The data acquisition module is used to collect bio-eye measurement parameters, intraocular lens constant, intraocular lens refractive power, and actual spherical equivalent data after cataract surgery from multiple patients. The residual data acquisition module calculates the baseline predicted spherical equivalent data for each patient based on the SRK / T formula, and acquires the spherical equivalent residual data based on the actual spherical equivalent data and the baseline predicted spherical equivalent data. The sample set construction module is used to construct a sample set based on the bio-eye measurement parameters, intraocular lens constant, intraocular lens refractive power, and spherical equivalent residual data of the multiple patients; The residual prediction model acquisition module is used to train the multilayer perceptron regression model by taking the biological eye measurement parameters, intraocular lens constant, and intraocular lens refractive power of the sample set as input features and the spherical equivalent residual data as output features, to obtain the spherical equivalent residual prediction model. The residual prediction module is used to input the target biological eye measurement parameters, target intraocular lens constant, and target intraocular lens refractive power of the target patient into the spherical equivalent residual prediction model to obtain the spherical equivalent prediction residual value. The refractive prediction result acquisition module is used to obtain the intraocular lens refractive prediction result of the target patient based on the target patient's baseline predicted spherical equivalent data and the spherical equivalent prediction residual value.
9. The intraocular lens refractive prediction device according to claim 8, characterized in that, The biological eye measurement parameters include axial length, corneal curvature, anterior chamber depth, corneal thickness, lens thickness, and corneal diameter.
10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by the processor, it implements the steps of the intraocular lens refractive prediction method as described in any one of claims 1 to 7.