Method and system for judging suitability of icl crystal implantation, electronic device and storage medium

Abstract

The application discloses an ICL crystal implantation adaptability judgment method and system, an electronic device and a storage medium, and relates to the technical field of medical treatment. The judgment method comprises the following steps: collecting eye parameter data of a patient; generating a classification model through a multi-parameter fusion analysis algorithm by using the eye parameter data; inputting the eye parameter data of a patient to be diagnosed into the classification model, and outputting an adaptability judgment result, wherein the judgment result is a first result or a second result. By integrating the eye anatomical structure parameters and the physiological function parameters, and combining the improved weighted fuzzy support vector machine (WFSVM) algorithm, the limitations of traditional single parameter evaluation are solved, and the adaptability judgment accuracy of critical cases is significantly improved, thereby avoiding the surgical risks caused by misdiagnosis.

Description

Technical Field

[0001] The present invention relates to the field of ophthalmic medicine, and in particular to an ICL lens implant adaptability judgment method, system, electronic device and storage medium. Background Art

[0002] As people's demand for vision correction continues to increase, ICL lens implantation surgery, as an effective means of vision correction, is becoming more and more widely used. However, there are still some shortcomings in the diagnostic evaluation process before ICL lens implantation surgery. For some special eye conditions, such as small eyeballs, combined risk of keratoconus, short ciliary processes, thickened lenses, etc., the existing judgment methods are difficult to accurately determine whether they are suitable for ICL surgery. In addition, there is a lack of intelligent decision-making support in the choice of surgical method, including incision direction, lens placement, lens size, etc., which may lead to poor surgical results or the occurrence of postoperative complications. At the same time, the existing methods for predicting postoperative vision and assessing the risk of postoperative complications also have certain limitations and cannot provide accurate reference for patients and doctors. Summary of the Invention

[0003] In order to overcome the defects in the prior art, the first purpose of the present invention is to provide a method for judging the suitability of ICL lens implantation, the second purpose of the present invention is to provide an ICL lens implantation suitability judgment system, the third purpose of the present invention is to provide an electronic device, and the fourth purpose of the present invention is to provide a computer-readable storage medium.

[0004] In order to achieve the above object, the technical solution adopted by the present invention is:

[0005] In a first aspect, a method for determining suitability for ICL lens implantation comprises the following steps:

[0006] Collect the patient's eye parameter data;

[0007] Using eye parameter data, a classification model is generated through a multi-parameter fusion analysis algorithm;

[0008] The eye parameter data of the patient to be diagnosed is input into the classification model, and the adaptability judgment result is output, and the judgment result includes a first result and a second result.

[0009] Optionally, the suitability determination result is a binary classification result, including a first result indicating suitability for surgery and a second result indicating unsuitability for surgery. Depending on application requirements, the suitability determination result can also be set to a related numerical value for further judgment by the user.

[0010] Optionally, the multi-parameter fusion analysis algorithm includes:

[0011] Normalizing the eye parameter data to map each parameter value to a uniform numerical range;

[0012] Based on the normalized data, the comprehensive weight of each parameter is calculated using information entropy and preset scoring values;

[0013] Mapping the normalized parameter value to a suitability membership through a fuzzy membership function, wherein the suitability membership is used to quantify the probability that the patient meets the first result;

[0014] Performing weighted fusion on the adaptability membership based on the comprehensive weight to generate a weighted feature vector, which is divided into a training set and a validation set;

[0015] Using the training set and the corresponding adaptability labels, the weighted fuzzy support vector machine classifier is trained to generate an initial model;

[0016] The initial model is verified using the validation set and the corresponding adaptability labels until the preset performance indicators are reached and the classification model is obtained.

[0017] By normalizing ocular parameters to eliminate dimensional differences, combining information entropy quantitative analysis with preset scoring values ​​to dynamically allocate parameter weights, and using fuzzy membership functions to accurately map the adaptation probability of critical parameter values, a classification model based on weighted fuzzy support vector machines is constructed. After clinical data verification and reaching the preset performance indicators, a binary judgment result of "first result" or "second result" is output, which effectively solves the problem that traditional methods rely on subjective experience and have a high misjudgment rate for borderline cases, significantly improving the objectivity and accuracy of adaptive diagnosis, and is particularly suitable for the rapid and accurate evaluation of complex cases.

[0018] Optionally, the eye parameters include anatomical parameters and physiological function parameters, the anatomical parameters include at least one of corneal diameter, corneal curvature, corneal thickness, anterior chamber depth, anterior chamber volume, chamber angle, iris morphology, ciliary process position and morphology, lens morphology, vitreous cavity depth, and axial length; the physiological function parameters include at least one of intraocular pressure, corneal endothelial cell count, tear film stability, and retinal blood flow. Preferably, the anatomical parameters and physiological function parameters each include at least three. By comprehensively evaluating the anatomical parameters and physiological function parameters of the eye, the comprehensiveness and reliability of the ICL lens implant adaptation judgment are significantly improved.

[0019] The anatomical parameters can be measured using high-precision imaging equipment, such as optical coherence tomography (OCT) and ultrasound biomicroscopy (UBM), to obtain detailed anatomical parameters such as corneal diameter, corneal curvature, corneal thickness, anterior chamber depth, anterior chamber volume, chamber angle, iris morphology, ciliary process position and morphology, lens morphology, vitreous cavity depth, and axial length. Specifically, each parameter can be measured using the following equipment:

[0020] Corneal diameter: white-to-white distance (WTW, HWTW) was obtained using corneal topography, OCT, UBM, or intraocular lens biometry (IOL-master);

[0021] Corneal curvature (K): Scheimpflug imaging (such as Pentacam) or anterior segment OCT is used to obtain the curvature radius of the anterior and posterior corneal surfaces, and the mean curvature (Km) and steepest curvature (Kmax) are calculated;

[0022] Corneal thickness (CCT): This method measures the central corneal thickness using corneal topography or optical coherence tomography (OCT). It can also measure the thickness distribution of different areas of the cornea, such as the thinnest point thickness (TPT).

[0023] Anterior chamber depth (ACD): The distance from the corneal endothelium to the anterior surface of the lens is measured using OCT or UBM, which can distinguish between central anterior chamber depth (CACD) and peripheral anterior chamber depth (PACD);

[0024] Anterior chamber volume (AVC): measured using corneal topography, optical coherence tomography (OCT), or Scheimpflug imaging techniques (such as Pentacam);

[0025] Anterior chamber angle (ACA): directly observed or measured by imaging techniques such as ultrasound biomicroscopy (UBM) or anterior segment optical coherence tomography (AS-OCT);

[0026] Iris morphology: Use OCT or UBM to obtain parameters such as the iridocorneal angle (ICA) and irido-ciliary epithelial cysts;

[0027] Ciliary process (CP) position and morphology: UBM is used to obtain ciliary sulcus to ciliary sulcus distance (STS), ciliary sulcus morphology (MCS), ciliary sulcus width (LD-ITC), etc.

[0028] Lens morphology: UBM, OCT or A-ultrasound were used to measure lens thickness (LT), sagittal height of the anterior lens surface (STSL), and zonules of the lens;

[0029] Vitreous cavity depth (VCD): The distance from the posterior surface of the lens to the inner limiting membrane of the retina is measured using A-ultrasound;

[0030] Axial length (AL): measured using an optical or ultrasonic biometer.

[0031] Specifically, each physiological function parameter can be measured using the following equipment:

[0032] Intraocular pressure (IOP): measured using a non-contact tonometer or Goldmann applanation tonometer;

[0033] Corneal endothelial cell count (ECD): Count the number of endothelial cells per unit area using a corneal endothelial microscope;

[0034] Tear film stability: measurement of tear breakup time (BUT).

[0035] Retinal blood flow: Retinal blood flow density was assessed using optical coherence tomography angiography (OCTA).

[0036] The normalization process refers to mapping the eye parameters of different dimensions to a uniform interval [0,1] through linear changes. For example, for the parameter , calculate its normalized value ,in and where _ ...

[0037] Optionally, the calculation of the comprehensive weight of each parameter by using information entropy and a preset score value includes:

[0038] According to the collected eye parameter data, the information entropy of each parameter is calculated ,in H ( x i ) =− Σ[p(x ij )*log2(p(x ij ))] , is the probability of the jth value of the i-th parameter occurring;

[0039] Based on preset scoring values ​​in historical clinical datasets , binding regulatory factors Information entropy and the preset rating value Perform weighted calculation to obtain comprehensive weight ,in , is the maximum value of the information entropy of all parameters, is a balance coefficient between 0 and 1, The range is 0 to 1. The smaller the information entropy value, the more concentrated the distribution of the feature value, the greater the amount of information, and the higher the weight should be.

[0040] The preset score value is a predefined score of eye parameter importance based on historical clinical data sets and medical consensus, which is used to quantify the contribution of different parameters to the judgment of ICL surgical suitability. The preset score value ranges from 0 to 1. Specifically, it can be determined by statistically analyzing the correlation between various parameters (such as corneal curvature, anterior chamber depth, ciliary sulcus distance, etc.) and surgical suitability in thousands of successful surgical cases. Feature importance analysis of historical data is performed using algorithms such as logistic regression or random forest, and is verified by clinical medical guidelines to ensure its scientificity and repeatability. The historical clinical data set refers to a collection of standardized medical data accumulated in clinical practice related to ICL surgery.

[0041] Optionally, mapping the normalized parameter value to a suitability membership through a fuzzy membership function, where the suitability membership is used to quantify the probability that the patient meets the first result, includes:

[0042] For each normalized eye parameter value, define a fuzzy membership function belonging to the first result, wherein the fuzzy membership function is a piecewise linear function including a first threshold interval, a second threshold interval, and a third threshold interval;

[0043] When the eye parameter value is in the first threshold interval, the membership degree is 0;

[0044] When the eye parameter value is in the second threshold interval, the membership degree increases linearly with the eye parameter value;

[0045] When the eye parameter value is in the third threshold interval, the membership degree is 1;

[0046] The division of the threshold interval is based on the statistical analysis results of the corresponding parameters in the historical clinical data set.

[0047] Specifically, the piecewise linear function may be a triangular function or a trapezoidal function. Preferably, as the amount of data in the historical clinical data set increases, the threshold and the function shape are gradually adjusted.

[0048] For example, the fuzzy membership of the anterior depth (ACD) can be set as follows: based on the historical clinical dataset and medical consensus, the membership function of the anterior depth is designed using a piecewise linear design.

[0049] When the ACD is less than 2.8 mm, the patient is judged to be completely unsuitable for surgery, and the membership degree is set to 0;

[0050] When the ACD is between 2.8 mm and 3.2 mm, the membership increases linearly with depth, and the calculation formula is: , reflecting the gradual improvement of adaptability within this range;

[0051] When the ACD exceeds 3.2 mm, the membership reaches a maximum value of 1, indicating that the area is completely suitable for surgery.

[0052] For example, if the patient's ACD is 3.0 mm, the membership is (3.0-2.8) / 0.4=0.5, which is 50% fit. Meanwhile, the membership of "not suitable for surgery" is obtained by subtracting the fit from 1, ensuring logical complementarity.

[0053] After completing the fuzzy membership mapping of each parameter, the algorithm is based on the pre-calculated comprehensive weight ( ) performs weighted fusion on the membership of “suitable for surgery”.

[0054] Subsequently, the weighted feature vector dataset was divided into a training set (70%-80%) and a validation set (20%-30%) using random stratified sampling to ensure a balanced distribution of the two types of samples (suitable / unsuitable for surgery) during training and validation. The training set was used to construct an initial model for the weighted fuzzy support vector machine (WFSVM) classifier, while the validation set was used to evaluate the model's classification performance, including accuracy, sensitivity, and specificity, through cross-validation. Based on the validation results, model parameters, such as the penalty factor C and kernel function parameters, were optimized until pre-defined performance metrics, such as an accuracy of ≥95%, were achieved.

[0055] After the weighted feature vector is constructed, the multi-parameter fusion analysis algorithm uses the training set and its corresponding suitability label ("suitable for surgery" or "not suitable for surgery") to train the weighted fuzzy support vector machine (WFSVM) classifier. The specific process is as follows:

[0056] (1) Model initialization

[0057] The objective function of WFSVM is defined as: ;

[0058] The constraints are: ;

[0059] in, is the normal vector of the hyperplane;

[0060] is the bias term;

[0061] is a penalty factor used to balance the margin maximization and classification error;

[0062] It is a slack variable that allows some samples to be misclassified;

[0063] It is The weight of each feature;

[0064] It is The fuzzy membership of samples (for positive samples, ; For negative samples, );

[0065] is a kernel function that maps the input features to a high-dimensional space (for example, a radial basis function RBF kernel can be used);

[0066] is the label of the sample (+1 for fit, -1 for not fit).

[0067] (2) Model training

[0068] The weighted feature vectors of the training set are input into WFSVM, and the objective function is solved through an iterative optimization algorithm to find the optimal hyperplane that maximizes the classification interval and minimizes the weighted classification error.

[0069] The kernel function parameters and penalty factor C are adjusted according to clinical data to balance model complexity and generalization ability.

[0070] (3) Model validation

[0071] The initial model was cross-validated using the validation set to evaluate classification performance indicators, including accuracy, sensitivity (positive class recognition rate), and specificity (negative class recognition rate).

[0072] If the model performance does not reach the preset threshold (such as accuracy ≥ 95%), the parameters are readjusted or the feature weights are optimized and retrained until the requirements are met to obtain a classification model.

[0073] After obtaining the classification model, the eye parameter data of the patient to be diagnosed is input into the classification model to obtain a judgment result on whether the patient is suitable for ICL lens implantation surgery.

[0074] The suitability label is a predefined classification identifier based on the actual surgical outcomes of patients in historical clinical datasets and medical consensus, which is used to supervise the training process of the machine learning model. The label is a binary variable, including suitable for surgery and unsuitable for surgery.

[0075] Optionally, the judgment method further includes, when the judgment result is the first result, generating a surgical plan using a rules engine. The surgical plan includes incision direction, lens placement, and lens size. The rules engine is an automated decision-making system built based on predefined medical rules and historical clinical datasets. Its core consists of a rule base and a reasoning mechanism. The rule base integrates medical standards for ICL surgical suitability judgment and plan generation. These medical standards can be referenced in documents such as the "Clinical Expert Consensus on ICL Surgery (2018)", "Chinese Expert Consensus on Posterior Chamber Intraocular Lens Implantation in Phakic Eyes (2019)", and "Chinese Expert Consensus on Clinical Expert Consensus on Phakic Intraocular Lens Implantation in Phakic Eyes (2023 Edition)". Conditional statements (IF-THEN rules) are used to link patient ocular parameters with surgical procedure requirements. The reasoning mechanism dynamically matches the conditions in the rule base based on input parameters (such as ACD, STS, corneal curvature, etc.) to output a surgical plan that meets clinical standards. The surgical plan includes detailed information on incision direction, lens placement, and lens size. Specifically, the judgment logic of the rules engine can be determined as follows:

[0076] 1. Incision direction:

[0077] Steep meridian: The direction of the steepest meridian is calculated based on corneal topography;

[0078] Astigmatism axis: Consider the patient's astigmatism axis and try to choose a direction perpendicular or nearly perpendicular to the astigmatism axis to reduce postoperative residual astigmatism;

[0079] Avoid vital structures: Avoid areas of corneal degeneration.

[0080] 2. Crystal placement:

[0081] Horizontal implantation: For most patients, horizontal implantation (0 or 180 degrees) is preferred;

[0082] Vertical implantation: For patients with mild astigmatism who do not need to implant an astigmatism lens, vertical implantation (90 degrees) can be considered;

[0083] Oblique axial implantation: For patients with significant corneal astigmatism who need to correct astigmatism, oblique axial implantation can be selected based on the axis and size of the astigmatism;

[0084] Avoid vital structures: Avoid iris cysts.

[0085] 3. Crystal size selection:

[0086] WTW (White-to-White) measurement: Horizontal corneal diameter (WTW) was measured using corneal topography or anterior segment OCT;

[0087] Sulcus-to-Sulcus (STS) measurement: UBM was used to measure the distance from ciliary sulcus to ciliary sulcus (STS);

[0088] Crystal size calculation formula:

[0089] ICL size = STS + 0.5 mm (based on the STS method);

[0090] Or ICL size = WTW + (0.5-1.0) mm (based on the WTW method, it needs to be adjusted according to the anterior chamber depth and ciliary process morphology). Attention should be paid to the arch height to prevent it from being too high or too low.

[0091] Optionally, the judgment method further includes, after generating the surgical plan, performing risk prediction on the surgical plan based on a risk prediction model, and adjusting the surgical plan according to the prediction results to obtain an optimized plan; the risk prediction model is a multivariate logistic regression model, the input parameters of which are the patient's anatomical structure parameters, physiological function parameters, and surgical plan parameters, and the probability of postoperative complications is calculated using a logistic function. Specifically, the risk prediction model expression is: ,in As risk factors, are the regression coefficients learned from the historical clinical dataset via maximum likelihood estimation.

[0092] Preferably, the risk can be stratified according to the predicted probability value, such as low risk ( ), medium risk ( ) and high risk ( For patients with medium or high risk, the system automatically adjusts the surgical plan to reduce risk, such as selecting a smaller ICL, changing the incision location, and strengthening postoperative follow-up. The optimized plan will be revalidated based on the patient's individual health status (such as age and systemic diseases) to ensure that the adjusted plan meets clinical safety standards.

[0093] Optionally, the judgment method further includes, after obtaining the optimization plan, predicting postoperative vision through a postoperative prediction model; the postoperative prediction model is an artificial neural network model.

[0094] This postoperative prediction model takes as input patient parameters such as preoperative refraction, axial length (AL), anterior chamber depth (ACD), corneal curvature, lens size, and surgical procedure. Using a multilayer perceptron (MLP) or radial basis function network (RBFN) architecture, it models nonlinear relationships and outputs a predicted value for postoperative uncorrected visual acuity (UCVA). The model is trained on a historical clinical dataset encompassing preoperative parameters, surgical details, and postoperative visual recovery data from thousands of successful surgeries. The network weights are optimized using a backpropagation algorithm, and activation functions such as sigmoid or ReLU are employed to enhance the model's nonlinear fitting capabilities. During the prediction process, the system combines key parameters from the optimized solution (such as lens size and incision orientation) with the patient's ocular anatomical characteristics (such as corneal thickness and vitreous cavity depth) to simulate the postoperative imaging effects of the optical system and quantitatively assess the level of visual recovery. For example, for patients with high myopia and retinal blood flow abnormalities, the model can combine lens optical design parameters and retinal thickness distribution to more accurately predict postoperative visual recovery, providing patients with reasonable expectations.

[0095] Optionally, the judgment method further includes generating a risk assessment report through a complication risk assessment system after the operation.

[0096] The complication sorting and assessment system integrates preoperative ocular parameters (such as anterior chamber depth, lens thickness, and axial length), surgical details (such as incision direction and lens size), and early postoperative monitoring data (such as intraocular pressure fluctuations and changes in corneal endothelial cell density). It uses statistical methods based on multivariate logistic regression models and big data analysis to assess the probability of common complications such as infection, high intraocular pressure, cataracts, and retinal detachment, thereby providing early warning of possible complication risks after surgery. For example, for some patients with large postoperative intraocular pressure fluctuations, the risk of high intraocular pressure complications can be promptly assessed by combining factors such as their preoperative intraocular pressure level, ocular anatomy, and the degree of interference with intraocular tissue during surgery, and appropriate preventive and treatment measures can be taken.

[0097] In a second aspect, a system for determining suitability of ICL lens implantation includes:

[0098] A data acquisition module, used to collect the patient's eye parameter data;

[0099] A multi-parameter fusion analysis module is used to normalize the eye parameter data and map each parameter value to a unified numerical range; based on the normalized data, calculate the comprehensive weight of each parameter through information entropy and a preset score value; map the normalized parameter value to an adaptability membership through a fuzzy membership function; perform weighted fusion on the adaptability membership based on the comprehensive weight to generate a weighted feature vector, which is divided into a training set and a validation set; use the training set and the corresponding adaptability label to train a weighted fuzzy support vector machine classifier to generate an initial model; and use the validation set and the corresponding adaptability label to validate the initial model until the preset performance index is achieved, thereby obtaining a classification model.

[0100] an output module, configured to input the eye parameter data of the patient to be diagnosed into the classification model and output a suitability judgment result, wherein the judgment result is a first result or a second result;

[0101] A rule engine module, configured to generate a surgical plan through a rule engine when the judgment result is the first result;

[0102] A risk prediction module is used to predict the risk of the surgical plan based on the risk prediction model after the surgical plan is generated, and to adjust the surgical plan according to the prediction results to obtain an optimized plan;

[0103] The postoperative prediction module is used to predict postoperative vision through the postoperative prediction model after obtaining the optimization plan, and to generate a risk assessment report through the complication risk assessment system after the operation.

[0104] In a third aspect, an electronic device includes a processor and a memory, wherein the memory stores a computer program, and when the program is executed by the processor, the above-mentioned judgment method is implemented.

[0105] In a fourth aspect, a computer-readable storage medium stores a computer program, which implements the above-mentioned judgment method when the program is executed.

[0106] Due to the application of the above technical solution, the present invention has the following advantages compared with the prior art:

[0107] 1. By integrating ocular anatomical parameters and physiological function parameters and combining them with an improved weighted fuzzy support vector machine (WFSVM) algorithm, the limitations of traditional single parameter evaluation are overcome. In particular, the accuracy of suitability judgment for critical cases is significantly improved, avoiding surgical risks caused by misdiagnosis.

[0108] 2. Based on a rules engine and multivariate logistic regression model, it dynamically generates personalized surgical plans (e.g., incision direction, lens size) and predicts the risk of postoperative complications (e.g., elevated intraocular pressure, cataracts). By pre-defining clinical data thresholds and implementing real-time optimization and adjustment, it reduces postoperative complications caused by improper surgical parameter selection, thereby improving surgical success rates and the patient's long-term visual quality.

[0109] 3. A postoperative vision prediction model constructed using an artificial neural network (ANN) combines multiple factors, including preoperative refraction, axial length, and lens optical design, to quantitatively predict postoperative uncorrected visual acuity (UCVA), helping physicians and patients establish reasonable expectations. Furthermore, a complication risk assessment system dynamically tracks postoperative parameters (such as intraocular pressure and corneal endothelial changes) to enable early warning and intervention, reducing the risk of irreversible damage.

[0110] 4. From data collection and suitability assessment to surgical plan generation, risk prediction, and postoperative management, we provide intelligent decision support throughout the entire process, reduce reliance on manual experience, and promote the standardization and scalability of ICL surgical operations.

[0111] 5. For special cases that are difficult to treat with traditional methods, such as microphthalmos, short ciliary processes, and thickened lenses, a multi-parameter fusion algorithm is used to accurately quantify the eye's spatial accommodation capacity, expand the applicable population for ICL surgery, and provide treatment possibilities for more patients with complex cases.

[0112] In order to make the above and other objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0113] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0114] Figure 1 4 is a flow chart of a judgment method in an embodiment of the present invention. DETAILED DESCRIPTION

[0115] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0116] Example 1: See Figure 1 As shown, a method for determining the suitability of ICL lens implantation includes the following steps:

[0117] S1. Collect the patient's eye parameter data.

[0118] The eye parameters include anatomical parameters and physiological function parameters. The anatomical parameters include at least one of corneal diameter, corneal curvature, corneal thickness, anterior chamber depth, anterior chamber volume, chamber angle, iris morphology, ciliary process position and morphology, lens morphology, vitreous cavity depth, and axial length; the physiological function parameters include at least one of intraocular pressure, corneal endothelial cell count, tear film stability, and retinal blood flow. Preferably, the anatomical parameters and physiological function parameters each include at least three. By comprehensively evaluating the anatomical parameters and physiological function parameters of the eye, the comprehensiveness and reliability of the ICL lens implant adaptation judgment are significantly improved.

[0119] The anatomical structure parameters can be obtained through high-precision imaging equipment, such as optical coherence tomography (OCT) technology and ultrasound biomicroscopy (UBM), to obtain detailed anatomical structure parameters such as corneal diameter, corneal curvature, corneal thickness, anterior chamber depth, anterior chamber volume, chamber angle, iris morphology, ciliary process position and morphology, lens morphology, vitreous cavity depth, and axial length.

[0120] In the measurement of physiological function parameters, each parameter is detected by the following equipment: intraocular pressure (IOP) is measured using a non-contact tonometer or a Goldmann applanation tonometer; corneal endothelial cell count (ECD) requires accurate counting of the number of endothelial cells per unit area using a corneal endothelial microscope; tear film stability is assessed by measuring the tear breakup time (BUT); and retinal blood flow density is non-invasively assessed using optical coherence tomography angiography (OCTA) technology to obtain retinal microvascular blood flow status data.

[0121] S2. Using eye parameter data, a classification model is generated through a multi-parameter fusion analysis algorithm.

[0122] This step includes:

[0123] S21 . Normalize the eye parameter data and map each parameter value to a unified numerical range.

[0124] Normalization involves linearly mapping ocular parameters of varying dimensions to a uniform interval [0, 1]. Because the measurement units and numerical ranges of these parameters vary significantly, normalization can eliminate these discrepancies and prevent large parameters from receiving undue weight during model training. By compressing data to a fixed interval, it improves the convergence speed and stability of classifiers such as the weighted fuzzy support vector machine (WFSVM) while also reducing classification bias caused by uneven parameter distribution.

[0125] S22. Based on the normalized data, the comprehensive weight of each parameter is calculated using information entropy and preset scoring values.

[0126] Specifically, based on the collected eye parameter data, the information entropy of each parameter is calculated. ,in H ( x i ) =− Σ[p(x ij )*log2(p(x ij ))] , is the probability of the jth value of the i-th parameter occurring;

[0127] Based on preset scoring values ​​in historical clinical datasets , binding regulatory factors Information entropy and the preset rating value Perform weighted calculation to obtain comprehensive weight ,in , is the maximum value of the information entropy of all parameters, is a balance coefficient between 0 and 1, The range is 0 to 1. The smaller the information entropy value, the more concentrated the distribution of the feature value, the greater the amount of information, and the higher the weight should be.

[0128] S23. Mapping the normalized parameter value to an adaptability membership through a fuzzy membership function.

[0129] For each normalized eye parameter value, a fuzzy membership function belonging to the first result is defined, where the fuzzy membership function is a piecewise linear function including a first threshold interval, a second threshold interval, and a third threshold interval.

[0130] When the eye parameter value is in the first threshold interval, the membership degree is 0;

[0131] When the eye parameter value is in the second threshold interval, the membership degree increases linearly with the eye parameter value;

[0132] When the eye parameter value is in the third threshold interval, the membership degree is 1;

[0133] The division of the threshold interval is based on the statistical analysis results of the corresponding parameters in the historical clinical data set.

[0134] S24. Perform weighted fusion on the adaptability membership based on the comprehensive weight to generate a weighted feature vector, which is divided into a training set and a validation set.

[0135] Specifically, the membership degree corresponding to each parameter ( ) will be combined with its weight ( ) are multiplied together to generate weighted eigenvalues ​​that reflect the importance of the parameters and the degree of fit. The weighted eigenvalues ​​of all parameters are combined to form a multidimensional weighted eigenvector, which is used to characterize the overall fit of the patient's eye.

[0136] S25. Using the training set and the corresponding adaptability labels, train the weighted fuzzy support vector machine classifier to generate an initial model.

[0137] S26. Use the validation set and the corresponding adaptability labels to validate the initial model until the preset performance indicators are reached to obtain a classification model.

[0138] S3. Input the eye parameter data of the patient to be diagnosed into the classification model, and output the adaptability judgment result, which is the first result or the second result.

[0139] S4. When the judgment result is the first result, a surgical plan is generated through the rule engine.

[0140] The surgical plan includes incision direction, lens placement, and lens size. The rule engine is an automated decision-making system built based on predefined medical rules and historical clinical data. Its core consists of a rule base and a reasoning mechanism. The rule base integrates medical standards for ICL surgical suitability assessment and plan generation, linking patient ocular parameters with surgical procedure requirements through conditional statements (IF-THEN rules). The reasoning mechanism dynamically matches the conditions in the rule base based on input parameters (such as ACD, STS, corneal curvature, etc.) to output a surgical plan that meets clinical standards. The surgical plan includes detailed information on incision direction, lens placement, and lens size.

[0141] S5. After the surgical plan is generated, the risk of the surgical plan is predicted based on the risk prediction model, and the surgical plan is adjusted according to the prediction results to obtain an optimized plan.

[0142] The risk prediction model is a multivariate logistic regression model. The input parameters are the patient's anatomical parameters, physiological function parameters, and surgical plan parameters. The probability of postoperative complications is calculated using a logistic function. Specifically, the risk prediction model expression is: ,in As risk factors, are the regression coefficients learned from the historical clinical dataset via maximum likelihood estimation.

[0143] In one possible embodiment, the risk level can be stratified based on the predicted probability value, including low risk, medium risk, and high risk. For medium and high risk patients, the system automatically adjusts the surgical plan to reduce the risk.

[0144] S6. After obtaining the optimized plan, the postoperative visual acuity is predicted using the postoperative prediction model.

[0145] The postoperative prediction model is an artificial neural network model. It uses the patient's preoperative refractive power, axial length (AL), anterior chamber depth (ACD), corneal curvature, lens size, and surgical method as inputs. It uses a multilayer perceptron (MLP) or radial basis function network (RBFN) structure to perform nonlinear relationship modeling and outputs a predicted value for postoperative uncorrected visual acuity (UCVA).

[0146] S7. After the operation, a risk assessment report is generated through the complication risk assessment system.

[0147] The complication sorting and evaluation system integrates preoperative ocular parameters (such as anterior chamber depth, lens thickness, and axial length), surgical operation details (such as incision direction and lens size), and early postoperative monitoring data (such as intraocular pressure fluctuations and changes in corneal endothelial cell density). It uses statistical methods based on multivariate logistic regression models and big data analysis to evaluate the probability of common complications such as infection, high intraocular pressure, cataracts, and retinal detachment, thereby providing early warning of possible complication risks after surgery.

[0148] This embodiment also discloses an ICL lens implant suitability determination system, comprising:

[0149] A data acquisition module, used to collect the patient's eye parameter data;

[0150] A multi-parameter fusion analysis module is used to normalize the eye parameter data and map each parameter value to a unified numerical range; based on the normalized data, calculate the comprehensive weight of each parameter through information entropy and a preset score value; map the normalized parameter value to an adaptability membership through a fuzzy membership function; perform weighted fusion on the adaptability membership based on the comprehensive weight to generate a weighted feature vector, which is divided into a training set and a validation set; use the training set and the corresponding adaptability label to train a weighted fuzzy support vector machine classifier to generate an initial model; and use the validation set and the corresponding adaptability label to validate the initial model until the preset performance index is achieved, thereby obtaining a classification model.

[0151] an output module, configured to input the eye parameter data of the patient to be diagnosed into the classification model and output a suitability judgment result, wherein the judgment result is a first result or a second result;

[0152] A rule engine module, configured to generate a surgical plan through a rule engine when the judgment result is the first result;

[0153] A risk prediction module is used to predict the risk of the surgical plan based on the risk prediction model after the surgical plan is generated, and to adjust the surgical plan according to the prediction results to obtain an optimized plan;

[0154] The postoperative prediction module is used to predict postoperative vision through the postoperative prediction model after obtaining the optimization plan, and to generate a risk assessment report through the complication risk assessment system after the operation.

[0155] This embodiment further discloses an electronic device, including a processor and a memory, wherein the memory stores a computer program, and when the program is executed by the processor, the above-mentioned determination method is implemented.

[0156] This embodiment further discloses a computer-readable storage medium storing a computer program. When the program is executed, the above-mentioned determination method is implemented.

[0157] Example 2:

[0158] Patient A was a 28-year-old male with severe myopia in both eyes, a diopter of -12.00 D, and mild astigmatism. Optical coherence tomography and ultrasound biomicroscopy revealed an anterior chamber depth of 3.0 mm, short ciliary processes, and normal lens thickness, but slightly below-normal corneal endothelial cell density. A multi-parameter fusion analysis algorithm, integrating parameters such as anterior chamber depth, ciliary process morphology, and corneal endothelial cell density, determined that Patient A was suitable for ICL implantation, but caution was required in the surgical plan. Based on Patient A's ocular features, the rule engine determined a surgical incision direction of 120 to 150 degrees temporally, with the lens positioned obliquely and a lens size of 12.1 to minimize pressure on ocular tissue. The postoperative visual acuity prediction model predicted that Patient A's postoperative visual acuity would be above 0.8, but close monitoring of corneal endothelial cell changes was necessary to prevent possible complications such as corneal edema.

[0159] Example 3:

[0160] Patient B is a 35-year-old female. Due to her high myopia and thin cornea, she is not suitable for laser myopia surgery and is considering ICL lens implantation. Examination found that her axial length is long, but she is at early risk of keratoconus. Corneal topography shows that the central area of ​​the cornea is thinning and protruding significantly. Through the multi-parameter fusion analysis of the present invention, combined with the corneal thickness distribution, corneal endothelial cell function and other anatomical structure parameters of the eye, it is believed that there are certain risks in patient B undergoing ICL surgery, which may aggravate corneal lesions. Therefore, it is recommended that patient B postpone the surgery and undergo corneal cross-linking and other treatments first, and then re-evaluate whether she is suitable for ICL surgery after the keratoconus condition stabilizes.

[0161] The ICL lens implantation suitability judgment method of the present invention can effectively solve the problems existing in the prior art, improve the accuracy and safety of ICL lens implantation surgery, and provide patients with more personalized and intelligent medical services.

[0162] Specific embodiments are used in the present invention to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core ideas. At the same time, for those skilled in the art, according to the ideas of the present invention, there may be changes in the specific implementation methods and application scopes. In summary, the contents of this specification should not be understood as limiting the present invention.

Claims

1. A method for determining suitability of ICL lens implantation, characterized in that: The following steps are involved: Collect the patient's eye parameter data; Normalizing the eye parameter data to map each parameter value to a uniform numerical range; Based on the normalized data, the comprehensive weight of each parameter is calculated by information entropy and preset scoring values, including: calculating the information entropy of each parameter based on the collected eye parameter data ,in , The probability of the jth value of the i-th parameter occurring; based on the preset score value in the historical clinical data set , binding regulatory factors Information entropy and the preset rating value Perform weighted calculation to obtain comprehensive weight ,in , is the maximum value of the information entropy of all parameters, is a balance coefficient between 0 and 1; The normalized parameter values ​​are mapped to adaptive memberships through a fuzzy membership function, where the adaptive membership is used to quantify the probability that the patient meets the first result, including: for each normalized eye parameter value, defining a fuzzy membership function that belongs to the first result, where the fuzzy membership function is a piecewise linear function and includes first, second, and third threshold intervals; when the eye parameter value is within the first threshold interval, the membership is 0; when the eye parameter value is within the second threshold interval, the membership increases linearly with the eye parameter value; when the eye parameter value is within the third threshold interval, the membership is 1; wherein the division of the threshold intervals is based on the statistical analysis results of the corresponding parameters in the historical clinical data set; Performing weighted fusion on the adaptability membership based on the comprehensive weight to generate a weighted feature vector, which is divided into a training set and a validation set; Using the training set and the corresponding adaptability labels, the weighted fuzzy support vector machine classifier is trained to generate an initial model; The initial model is verified using the validation set and the corresponding fitness labels until the preset performance indicators are reached and a classification model is obtained; The eye parameter data of the patient to be diagnosed is input into the classification model, and an adaptability judgment result is output, and the judgment result is the first result or the second result.

2. The judgment method according to claim 1, characterized in that: The eye parameters include anatomical parameters and physiological function parameters. The anatomical parameters include at least one of corneal diameter, corneal curvature, corneal thickness, anterior chamber depth, anterior chamber volume, chamber angle, iris morphology, ciliary process position and morphology, lens morphology, vitreous cavity depth, and axial length; the physiological function parameters include at least one of intraocular pressure, corneal endothelial cell count, tear film stability, and retinal blood flow.

3. The judgment method according to claim 1, characterized in that: The judgment method also includes, when the judgment result is the first result, generating a surgical plan through a rule engine, the surgical plan including the incision direction, the lens placement position and the lens size; the rule engine is an operation database constructed based on pre-set medical rules and clinical data.

4. The judgment method according to claim 3, characterized in that: The judgment method also includes, after generating a surgical plan, performing risk prediction on the surgical plan based on a risk prediction model, and adjusting the surgical plan according to the prediction results to obtain an optimized plan; the risk prediction model is a multivariate logistic regression model.

5. The judgment method according to claim 4, characterized in that: The judgment method further includes, after obtaining the optimization plan, predicting postoperative vision using a postoperative prediction model; the postoperative prediction model is an artificial neural network model; Furthermore, after the operation, a risk assessment report is generated through the complication risk assessment system.

6. An ICL lens implant suitability judgment system, characterized in that: include: A data acquisition module, used to collect the patient's eye parameter data; A multi-parameter fusion analysis module is used to normalize the eye parameter data and map each parameter value to a unified numerical range; Based on the normalized data, the comprehensive weight of each parameter is calculated by information entropy and preset scoring values, including calculating the information entropy of each parameter based on the collected eye parameter data. ,in , is the probability of the jth value of the i-th parameter occurring; Based on preset scoring values ​​in historical clinical datasets , binding regulatory factors Information entropy and the preset rating value Perform weighted calculation to obtain comprehensive weight ,in , is the maximum value of the information entropy of all parameters, is a balance coefficient between 0 and 1; mapping the normalized parameter value to an adaptive membership through a fuzzy membership function, including defining, for each normalized eye parameter value, a fuzzy membership function belonging to the first result, wherein the fuzzy membership function is a piecewise linear function including first, second, and third threshold intervals; When the eye parameter value is in the first threshold interval, the membership is 0; when the eye parameter value is in the second threshold interval, the membership increases linearly with the eye parameter value; when the eye parameter value is in the third threshold interval, the membership is 1; wherein, the division of the threshold intervals is based on the statistical analysis results of the corresponding parameters in the historical clinical data set; based on the comprehensive weight, the adaptability membership is weighted and fused to generate a weighted feature vector, which is divided into a training set and a validation set; using the training set and the corresponding adaptability label, a weighted fuzzy support vector machine classifier is trained to generate an initial model; using the validation set and the corresponding adaptability label, the initial model is verified until the preset performance index is reached to obtain a classification model; an output module, configured to input the eye parameter data of the patient to be diagnosed into the classification model and output a suitability judgment result, wherein the judgment result is a first result or a second result; A rule engine module, configured to generate a surgical plan through a rule engine when the judgment result is the first result; A risk prediction module is used to predict the risk of the surgical plan based on the risk prediction model after the surgical plan is generated, and to adjust the surgical plan according to the prediction results to obtain an optimized plan; The postoperative prediction module is used to predict postoperative vision through the postoperative prediction model after obtaining the optimization plan, and to generate a risk assessment report through the complication risk assessment system after the operation.

7. An electronic device comprising a processor and a memory, characterized in that: The memory stores a computer program, and when the program is executed by the processor, the judgment method according to any one of claims 1 to 5 is implemented.

8. A computer-readable storage medium storing a computer program, characterized in that: When the program is executed, the judgment method according to any one of claims 1 to 5 is implemented.

Images