Intraocular lens recommendation method and apparatus, device, storage medium

Abstract

The embodiment of the application provides a kind of intraocular lens recommendation method and device, equipment, storage medium, belong to artificial intelligence technical field.The method comprises: according to the initial dialogue data of target object, initial eye measurement data and initial eye detection image carries out feature extraction, obtains initial object eye feature;Obtain the historical data of historical object library, historical data includes historical object data and historical intraocular lens category;According to initial object eye feature, historical object data is matched to determine target object data, and according to the historical intraocular lens category of target object data determines first predicted intraocular lens category;Obtain the candidate intraocular lens category of lens recommendation library;According to initial object eye feature and first predicted intraocular lens category carries out lens category matching, determines target intraocular lens recommendation category.The embodiment of the application can improve the efficiency of intraocular lens recommendation, and improve the accuracy of intraocular lens recommendation.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to a method, apparatus, device, and storage medium for recommending intraocular lenses. Background Technology

[0002] An intraocular lens (IOL) is an implanted eye that replaces the cloudy, aging natural lens during cataract surgery to perform refractive functions. There are many types of IOLs, mainly classified according to their materials and design. IOL recommendation methods can assist physicians in selecting IOLs for patients. Currently, in practice, IOL selection primarily relies on physicians' personal experience and patient biometric results. However, this method is too dependent on physician experience, making it difficult to achieve standardized judgment criteria and thus hindering the attainment of homogeneous or optimized selection results. Furthermore, traditional models cannot generalize well to changes in IOL availability and the diversity of patient descriptions, requiring continuous retraining with new IOL data, which reduces the efficiency and accuracy of IOL recommendations. Summary of the Invention

[0003] The main objective of this application is to provide a method, apparatus, device, and storage medium for recommending intraocular lenses (IOLs), aiming to improve the efficiency and accuracy of IOL recommendations.

[0004] To achieve the above objectives, a first aspect of this application provides a method for recommending intraocular lenses, the method comprising:

[0005] Acquire initial data of the target object, including initial dialogue data, initial eye measurement data, and initial eye detection images;

[0006] Based on the initial dialogue data, the initial eye measurement data, and the initial eye detection image, object eye features are extracted to obtain initial object eye features;

[0007] Obtain a preset historical object library, which includes historical data, including historical object data and the corresponding historical intraocular lens categories;

[0008] Based on the initial object's eye features, the historical object data is matched with historical objects to determine the target object data, and based on the historical intraocular lens category corresponding to the target object data, the first predicted intraocular lens category is determined.

[0009] Obtain a preset lens recommendation library, which includes candidate artificial lens categories;

[0010] Based on the initial object's eye features and the first predicted intraocular lens category, the candidate intraocular lens categories are matched to determine the target intraocular lens recommendation category, which is used to represent the intraocular lens category recommended for the target object.

[0011] In some embodiments, the step of extracting object eye features based on the initial dialogue data, the initial eye measurement data, and the initial eye detection image to obtain initial object eye features includes:

[0012] Based on the initial dialogue data, dialogue features are extracted to obtain the initial dialogue features;

[0013] Based on the initial eye measurement data, measurement data is extracted to obtain the object's eye measurement features;

[0014] Image feature extraction is performed based on the initial eye detection image to obtain the object's eye detection features;

[0015] The initial object eye features are obtained by encoding the initial dialogue features, the object eye measurement features, and the object eye detection features using a preset object information encoder.

[0016] In some embodiments, the step of extracting dialogue features based on the initial dialogue data to obtain initial dialogue features includes:

[0017] Key data is extracted from the initial dialogue data to obtain key dialogue data;

[0018] The key dialogue data is filled in according to the preset dialogue template to obtain the initial dialogue table;

[0019] The initial dialogue table is subjected to table feature extraction to obtain the initial dialogue features.

[0020] In some embodiments, the step of extracting dialogue features based on the initial dialogue data to obtain initial dialogue features includes:

[0021] The initial dialogue data is segmented into dialogue questions to obtain dialogue sub-data;

[0022] Text features are extracted from the dialogue sub-data to obtain dialogue text sub-features;

[0023] The initial dialogue features are obtained by concatenating the sub-features of the dialogue text.

[0024] In some embodiments, the step of extracting image features from the initial eye detection image to obtain object eye detection features includes:

[0025] The initial eye detection image is segmented to obtain an eye detection sub-image and image position data of the eye detection sub-image;

[0026] Local image encoding is performed on the eye detection sub-image to obtain eye detection sub-features;

[0027] The eye detection sub-features are spliced ​​together based on the image location data to obtain the object's eye detection features.

[0028] In some embodiments, the historical object data includes historical dialogue data, historical eye measurement data, and historical eye detection images;

[0029] The step of performing historical object matching on the historical object data based on the initial object's eye features to determine the target object data includes:

[0030] Based on the historical dialogue data, the historical eye measurement data, and the historical eye detection images, object eye features are extracted to obtain historical object eye features;

[0031] The similarity between the initial object's eye features and the historical object's eye features is calculated to obtain the object's eye similarity value.

[0032] The historical data is sorted according to the similarity value of the object's eyes to obtain a historical data sequence;

[0033] The historical data sequence is selected based on a preset similarity threshold to obtain target data, and the target object data is determined based on the historical object data corresponding to the target data.

[0034] In some embodiments, the step of performing lens category matching on the candidate intraocular lens categories based on the initial object eye features and the first predicted intraocular lens category to determine the target intraocular lens recommendation category includes:

[0035] Lens attribute features are extracted from the first predicted intraocular lens category to obtain the predicted intraocular lens attribute features;

[0036] The initial object's eye features and the predicted intraocular lens attribute features are fused to obtain the target object's eye features;

[0037] Based on the eye features of the target object, the candidate artificial lens categories are matched to determine the recommended category of the target artificial lens.

[0038] In some embodiments, the step of performing lens category matching based on the eye features of the target object to determine the recommended category of the target intraocular lens includes:

[0039] Lens attribute features are extracted from the candidate artificial lens categories to obtain the candidate artificial lens attribute features;

[0040] Similarity calculations are performed on the eye features of the target object and the lens attribute features of the candidate artificial lens to obtain a lens matching degree value;

[0041] The target intraocular lens (IOL) recommendation category is determined from the candidate IOL categories based on the lens matching degree value.

[0042] In some embodiments, determining the target intraocular lens recommendation category from the candidate intraocular lens categories based on the lens matching value includes:

[0043] Obtain candidate evaluation data for the candidate lens category;

[0044] The lens matching degree value is matched and evaluated based on the candidate evaluation data to determine the target matching degree value;

[0045] The target intraocular lens (IOL) recommendation category is determined from the candidate IOL categories based on the target matching degree value.

[0046] In some embodiments, the step of performing a matching evaluation on the lens matching degree value based on the candidate evaluation data to determine the target matching degree value includes:

[0047] Feature extraction is performed on the candidate evaluation data to obtain candidate evaluation features;

[0048] The candidate evaluation features are normalized to obtain candidate evaluation values;

[0049] The target matching value is obtained by weighting the candidate evaluation value and the lens matching value.

[0050] To achieve the above objectives, a second aspect of this application provides an intraocular lens recommendation device, the device comprising:

[0051] The data acquisition module is used to acquire the initial dialogue data, initial eye measurement data, and initial eye detection images of the target object;

[0052] The feature extraction module is used to extract object eye features based on the initial dialogue data, the initial eye measurement data, and the initial eye detection image to obtain initial object eye features;

[0053] The object library acquisition module is used to acquire a preset historical object library, which includes historical data, including historical object data and the corresponding historical intraocular lens categories;

[0054] The object matching module is used to perform historical object matching on the historical object data based on the initial object eye features to determine the target object data, and to determine the first predicted intraocular lens category based on the historical intraocular lens category corresponding to the target object data.

[0055] The recommendation library acquisition module is used to acquire a preset lens recommendation library, which includes candidate artificial lens categories;

[0056] The lens matching module is used to perform lens category matching on the candidate intraocular lens categories based on the initial object's eye features and the first predicted intraocular lens category, and determine the target intraocular lens recommended category, wherein the target intraocular lens recommended category is used to represent the intraocular lens category recommended for the target object.

[0057] To achieve the above objectives, a third aspect of this application provides an electronic device, comprising:

[0058] At least one memory;

[0059] At least one processor;

[0060] At least one computer program;

[0061] The at least one computer program is stored in the at least one memory, and the at least one processor executes the at least one computer program to perform:

[0062] As described in the first aspect above.

[0063] To achieve the above objectives, a fourth aspect of the present application provides a computer-readable storage medium storing a computer program for causing a computer to perform the method described in the first aspect above.

[0064] This application proposes an intraocular lens (IOL) recommendation method, apparatus, device, and storage medium. The method involves acquiring initial dialogue data, initial eye measurement data, and initial eye detection images of a target object. Based on these data, initial eye features are extracted to obtain initial object eye features. These initial object eye features fully consider various information related to the target object. A preset historical object library is acquired, including historical data. This historical data includes historical object data and corresponding historical IOL categories. Historical object matching is performed on the historical object data based on the initial object eye features to determine target object data. A first predicted IOL category is then determined based on the corresponding historical IOL category. A preset lens recommendation library is acquired, including candidate IOL categories. Lens category matching is performed on the candidate IOL categories based on the initial object eye features and the first predicted IOL category to determine a target IOL recommendation category. This target IOL recommendation category represents the IOL category recommended for the target object. Therefore, the embodiments of this application can progressively match intraocular lens (IOL) categories based on a preset historical object library and lens recommendation library, thereby improving the accuracy of IOL recommendations. Furthermore, based on the historical object library and lens recommendation library of the external device, this application can better adapt to the iterative updates of new IOL data, avoiding the need for continuous retraining by combining new IOL data. Based on this, the embodiments of this application can effectively improve the efficiency and accuracy of IOL recommendations. Attached Figure Description

[0065] Figure 1 This is a flowchart of the intraocular lens recommendation method provided in the embodiments of this application;

[0066] Figure 2 yes Figure 1 The flowchart of step S120 in the middle;

[0067] Figure 3 yes Figure 2 The flowchart of step S210 in the process;

[0068] Figure 4 yes Figure 2 The flowchart of step S230 in the text;

[0069] Figure 5 This is a schematic diagram illustrating feature extraction of the initial data of the target object provided in an embodiment of this application;

[0070] Figure 6 yes Figure 1 The flowchart of step S140 in the middle;

[0071] Figure 7 yes Figure 1 The flowchart of step S160 in the middle;

[0072] Figure 8 yes Figure 7 The flowchart of step S730 in the process;

[0073] Figure 9 This is a complete schematic diagram of the intraocular lens recommendation method provided in the embodiments of this application;

[0074] Figure 10 This is a schematic diagram of the structure of the intraocular lens recommendation device provided in the embodiments of this application;

[0075] Figure 11 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0076] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0077] It should be noted that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0078] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0079] First, let's analyze some of the terms used in this application:

[0080] Artificial intelligence (AI) is a new branch of computer science that studies, develops, and applies theories, methods, technologies, and systems to simulate, extend, and expand human intelligence. It aims to understand the essence of intelligence and produce intelligent machines that can react in a way similar to human intelligence. Research in this field includes robotics, speech recognition, image recognition, natural language processing, and expert systems. AI can simulate the information processes of human consciousness and thought. Furthermore, AI utilizes digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceiving the environment, acquiring knowledge, and using that knowledge to achieve optimal results.

[0081] Refractive intraocular lenses (IOLs) are artificial lenses used to correct refractive errors in the eye. Refractive errors occur when the eye cannot focus light correctly, leading to blurred vision. Refractive IOLs are surgically implanted into the eye to replace the existing lens, thereby improving vision. These IOLs are custom-made to fit the patient's refractive error for optimal correction.

[0082] Transformer is a deep learning model based on an attention mechanism. Transformer models can be used for natural language processing tasks such as machine translation and language modeling. By processing input sequences through self-attention, Transformers can perform parallel computation and capture long-range dependencies, thus excelling at processing long sequences of data. Transformer models have been successfully applied in various fields, including computer vision, speech recognition, and medical image analysis.

[0083] An intraocular lens (IOL) is an implanted eye that replaces the cloudy, aging natural lens during cataract surgery to provide refractive power. There are many types of IOLs, primarily classified according to their materials and design. Furthermore, with the rapid advancements in IOL materials and designs, an increasing number of refractive IOLs are available, offering personalized options for clinicians and patients. Choosing the right IOL for cataract surgery involves considering numerous factors. Recommended methods for selecting IOLs can assist doctors in making this selection process for patients.

[0084] Currently, in practical applications, the selection of intraocular lenses (IOLs) mainly relies on physicians' personal experience and patients' biometric results (such as pupil size, refractive status, axial length, corneal diameter, and curvature). However, this approach is too dependent on physician experience and cannot achieve uniform judgment criteria, making it difficult to achieve homogenized or more optimized selection results. Furthermore, when faced with updates to IOLs and the diversity of patient descriptions, traditional models cannot generalize well, requiring continuous retraining with new data, which reduces the efficiency and accuracy of IOL recommendations.

[0085] Based on this, embodiments of this application provide a method, apparatus, device, and storage medium for recommending intraocular lenses, aiming to improve the efficiency and accuracy of intraocular lens recommendations.

[0086] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0087] Foundational technologies for artificial intelligence generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies mainly encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0088] The intraocular lens recommendation method provided in this application relates to the field of artificial intelligence technology. The recommendation method provided in this application can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, etc.; the server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application implementing the intraocular lens recommendation method, but is not limited to the above forms.

[0089] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0090] Please see Figure 1 , Figure 1 This is an optional flowchart of the intraocular lens recommendation method provided in the embodiments of this application. In some embodiments of this application, Figure 1 The specific methods may include, but are not limited to, steps S110 to S160.

[0091] Step S110: Obtain initial data of the target object, the initial data including initial dialogue data, initial eye measurement data and initial eye detection image;

[0092] Step S120: Extract object eye features based on initial dialogue data, initial eye measurement data and initial eye detection image to obtain initial object eye features;

[0093] Step S130: Obtain a preset historical object library. The historical object library includes historical data, which includes historical object data and the corresponding historical intraocular lens categories.

[0094] Step S140: Based on the initial object's eye features, perform historical object matching on the historical object data to determine the target object data, and determine the first predicted intraocular lens category based on the historical intraocular lens category corresponding to the target object data.

[0095] Step S150: Obtain a preset lens recommendation library, which includes candidate artificial lens categories;

[0096] Step S160: Match the candidate intraocular lens categories with the initial object's eye features and the first predicted intraocular lens category to determine the target intraocular lens recommendation category. The target intraocular lens recommendation category is used to represent the intraocular lens category recommended for the target object.

[0097] In steps S110 to S160 of this application, initial dialogue data, initial eye measurement data, and initial eye detection images of the target object are acquired. Based on these data, initial eye measurement data, and initial eye detection images, object eye features are extracted to obtain initial object eye features. These initial object eye features fully consider various information related to the target object. A preset historical object library is acquired, which includes historical data. The historical data includes historical object data and the corresponding historical intraocular lens (IOL) categories. Historical object matching is performed on the historical object data based on the initial object eye features to determine the target object data. Based on the corresponding historical IOL categories, a first predicted IOL category is determined. A preset lens recommendation library is acquired, which includes candidate IOL categories. Lens category matching is performed on the candidate IOL categories based on the initial object eye features and the first predicted IOL category to determine the target IOL recommendation category. This target IOL recommendation category is used to represent the IOL category recommended for the target object. Therefore, the embodiments of this application can progressively match intraocular lens (IOL) categories based on a preset historical object library and lens recommendation library, thereby improving the accuracy of IOL recommendations. Furthermore, based on the historical object library and lens recommendation library of the external device, this application can better adapt to the iterative updates of new IOL data, avoiding the need for continuous retraining by combining new IOL data. Based on this, the embodiments of this application can effectively improve the efficiency and accuracy of IOL recommendations.

[0098] In step S110 of some embodiments, the target object refers to the object for whom an intraocular lens (IOL) needs to be recommended. For example, the target object can be a patient. Initial data refers to data associated with the target object. Initial dialogue data refers to data obtained through interviews with the target object. Initial dialogue data includes dialogue questions and the target object's responses to the dialogue questions. Dialogue questions include information such as the target object's eye habits, postoperative visual acuity expectations, expected life uses, financial capacity, and preference for the IOL. Therefore, in practical applications, this initial dialogue data is equivalent to a target information table constructed based on the dialogue questions and the target object's responses to the dialogue questions.

[0099] It should be noted that initial ocular measurement data refers to data obtained through precise measurements of the target subject using a biometric instrument. Initial ocular measurement data includes visual acuity, intraocular pressure, axial length, corneal diameter and curvature, central anterior chamber depth, lens thickness, astigmatism, and other ocular parameters. Based on this initial ocular measurement data, it is helpful to assess the target subject's pupil size, refractive status, and whether there are any coexisting eye diseases, thereby improving the accuracy of recommending intraocular lenses.

[0100] It's important to note that initial eye examination images refer to images generated through medical instruments used to examine the target subject's eyes. These images include fundus photographs, corneal topography, and optical coherence tomography (OCT) images. Corneal topography involves measuring the corneal surface of the target subject using specialized instruments to obtain images of the corneal curvature, shape, and topography. This data can be used to assess corneal health, assist doctors in diagnosing corneal diseases, and plan corneal surgery. Fundus OCT images are cross-sectional images of eye tissues generated by scanning reflected light signals using high-resolution imaging technology. Fundus OCT images can display details of eye structures, including the retina, optic nerve, and choroid, and are used to assist in diagnosing eye diseases, monitoring disease progression, and planning eye surgery. Therefore, initial eye examination images can be used to assess corneal morphology and retinal function.

[0101] It should be noted that the initial data may also include other data about the target object, which is not specifically limited here, in order to improve the accuracy of intraocular lens recommendations.

[0102] Understandably, when selecting an intraocular lens (IOL), many factors need to be considered. By taking into account the target patient's initial dialogue data, initial eye measurement data, and initial eye examination images, a thorough understanding of the target patient's lifestyle, eye habits, the biological anatomical characteristics of the patient's eyes, binocular vision, and other information can be obtained. This can effectively improve the accuracy of IOL recommendations, assist doctors in selecting suitable IOLs for the target patient, and enable the target patient to achieve better surgical results and better visual outcomes based on the IOL used.

[0103] It should be noted that in various specific embodiments of this application, when processing data related to the identity or characteristics of the target object, such as initial dialogue data, initial eye measurement data, initial eye detection images, and historical object data, is required, the object's permission or consent will be obtained first. Furthermore, the collection, use, and processing of this data will comply with relevant laws, regulations, and standards. In addition, when embodiments of this application require obtaining sensitive personal information of the object, separate permission or consent from the object will be obtained through pop-ups or redirection to a confirmation page. Only after obtaining the object's separate permission or consent will the necessary object-related data for the proper functioning of the embodiments of this application be obtained.

[0104] In step S120 of some embodiments, the initial data of the target object is input into a pre-trained lens recommendation model, which includes a feature extraction sub-model, a historical object matching sub-model, and a lens category matching sub-model. Thus, based on the feature extraction sub-model, object eye features are extracted from the initial dialogue data, initial eye measurement data, and initial eye detection images to obtain initial object eye features. Initial object eye features refer to the features extracted after comprehensively considering multiple types of data and images of the target object. The feature extraction sub-model can be an encoding structure built based on Transformer. Therefore, by inputting multiple types of initial data corresponding to the target object into the model for intraocular lens recommendation, this application can fully consider various information about the target object and improve the accuracy of the recommendation.

[0105] It should be noted that during the training of the lens recommendation model, other parameters of the intraocular lens that the target object expects to use can be input at the same time, such as: intraocular lens material, intraocular lens shape design, intraocular lens fixation method, intraocular lens optical zone function, intraocular lens price, reasons for selection, etc.

[0106] It should be noted that if the initial data also includes other data of the target object, then the other data will be input into the feature extraction layer together to update the initial object's eye features, thereby enriching the information contained in the extracted initial object's eye features.

[0107] Please see Figure 2 , Figure 2 This is a specific flowchart of step S120 provided in the embodiments of this application. In some embodiments of this application, step S120 may specifically include, but is not limited to, steps S210 to S240.

[0108] Step S210: Extract dialogue features based on the initial dialogue data to obtain the initial dialogue features;

[0109] Step S220: Extract measurement data based on the initial eye measurement data to obtain the object's eye measurement features;

[0110] Step S230: Extract image features based on the initial eye detection image to obtain the object's eye detection features;

[0111] Step S240: Encode the initial dialogue features, object eye measurement features and object eye detection features according to the preset object information encoder to obtain the initial object eye features.

[0112] In step S210 of some embodiments, initial dialogue features are used to characterize features extracted from the initial dialogue data. Initial dialogue features include semantic information, sentiment, dialogue structure, and other characteristics from the initial dialogue data.

[0113] In one embodiment, see Figure 3 , Figure 3 This is a specific flowchart of step S210 provided in the embodiments of this application. In some embodiments of this application, step S210 may specifically include, but is not limited to, steps S310 to S330.

[0114] Step S310: Segment the initial dialogue data into dialogue questions to obtain dialogue sub-data;

[0115] Step S320: Extract text features from the dialogue sub-data to obtain dialogue text sub-features;

[0116] Step S330: Perform feature concatenation on the sub-features of the dialogue text to obtain the initial dialogue features.

[0117] In step S310 of some embodiments, dialogue sub-data refers to the data consisting of each dialogue question and its corresponding response data in the initial dialogue data. Dialogue question segmentation of the initial dialogue data involves dividing the entire dialogue data according to question boundaries to obtain individual question parts in the dialogue. These segmented question parts are the dialogue sub-data, which includes the question and its corresponding response. The specific process of dialogue question segmentation includes word segmentation of the initial dialogue data to obtain dialogue word segmentation data; sentence boundary recognition based on the dialogue word segmentation data to obtain recognized sentence data; and semantic analysis based on the recognized sentence data to determine the dialogue sub-data. Dialogue question segmentation helps the system better understand and process dialogues, allowing each question or dialogue segment to be processed and analyzed independently. For example, the initial dialogue data includes three dialogue questions (question 1, question 2, and question 3), and correspondingly, the target object's response data based on the dialogue questions also includes three responses (e.g., response 1, response 2, and response 3). At this point, the initial dialogue data is segmented into three dialogue sub-data: (Question 1, Response 1), (Question 2, Response 2), and (Question 3, Response 3).

[0118] In step S320 of some embodiments, text features are extracted for each dialogue sub-data to obtain corresponding dialogue text sub-features. If the initial dialogue data includes three dialogue sub-data, then three dialogue text sub-features can be obtained. The dialogue text sub-features can be the result of concatenating features corresponding to text keywords in the dialogue sub-data, or they can be the result of text encoding the entire dialogue sub-data; no specific limitation is made here.

[0119] In step S330 of some embodiments, after determining the dialogue text sub-features corresponding to each dialogue sub-data, the multiple dialogue text sub-features are concatenated to obtain the initial dialogue features corresponding to the initial dialogue data.

[0120] In another embodiment, step S210 may further include, but is not limited to, the following steps:

[0121] Key data is extracted from the initial dialogue data to obtain key dialogue data.

[0122] Fill in the key dialogue data according to the preset dialogue template to obtain the initial dialogue form;

[0123] The initial dialogue table is subjected to table feature extraction to obtain the initial dialogue features.

[0124] As we can understand, key dialogue data refers to the crucial information extracted from the initial dialogue data. Key dialogue data may include the question posed, the request, keywords used in the target audience's responses, and contextual information about those responses. Therefore, using key dialogue data can help the model focus on the most important parts of the dialogue, reduce unnecessary information and noise, and improve the efficiency and accuracy of subsequent processing.

[0125] It's understandable that a preset dialogue template refers to a pre-set template used to structure key data. The preset dialogue template can include the target audience's personal information, interests, preferences, etc. Based on the preset dialogue template, key dialogue data is filled into the template to generate an initial dialogue table, thus transforming the target audience's initial dialogue data into structured tabular data. Through structured dialogue tables, dialogue data can be more easily analyzed, processed, and subsequently applied, improving data operability and scalability.

[0126] Understandably, the initial dialogue features at this stage refer to the combination of features extracted from the initial dialogue table. These initial dialogue features can extract multi-level feature information about the dialogue, such as word frequency, dialogue length, and sentiment. By extracting features from the initial dialogue table, the model can better understand the characteristics and patterns of the dialogue data, providing a foundation for subsequent analysis, mining, and modeling.

[0127] In step S220 of some embodiments, the object's eye measurement features are used to characterize the features extracted from the initial eye measurement data. If the initial eye measurement data is text data, the process of extracting features from the text data is the same as the process of extracting initial dialogue features from the initial dialogue data described above, except that the initial dialogue data is replaced by the initial eye measurement data, which will not be repeated here.

[0128] In step S230 of some embodiments, object eye detection features are used to characterize features extracted from the initial eye detection image.

[0129] Please see Figure 4 , Figure 4 This is a specific flowchart of step S230 provided in the embodiments of this application. In some embodiments of this application, step S230 may specifically include, but is not limited to, steps S410 to S430.

[0130] Step S410: Perform image segmentation on the initial eye detection image to obtain eye detection sub-images and image position data of the eye detection sub-images;

[0131] Step S420: Perform local image encoding on the eye detection sub-image to obtain eye detection sub-features;

[0132] Step S430: Perform feature stitching on the eye detection sub-features based on the image position data to obtain the object's eye detection features.

[0133] In step S410 of some embodiments, the eye detection sub-image refers to the image after being segmented from the initial eye detection image. The segmentation size of the eye detection sub-image can be determined according to preset image segmentation parameters, and the image segmentation parameters can be flexibly set according to actual needs and the processing capability of the model, which will not be elaborated here. Image position data refers to the position coordinates, size, and other data of the eye detection sub-image in the original initial eye detection image. This position data will be used for subsequent feature stitching. Image segmentation of the initial eye detection image can be performed on the image by sliding window or by dividing it according to a predefined grid, so that each eye detection sub-image can be called a block (or patch).

[0134] In step S420 of some embodiments, the eye detection sub-features refer to the encoded features corresponding to each eye detection sub-image. The process of local image encoding of the eye detection sub-images can employ linear projection, where the eye detection sub-features are represented as features mapped to a new space. Linear projection refers to the process of mapping data to a new space using a linear transformation. Therefore, this application can treat each patch as a vector, and use linear projection (such as matrix multiplication) to map the vector representation of each patch to a new space, obtaining the mapped patch vectors. Then, these mapped patch vectors are treated as tokens, i.e., eye detection sub-features. Based on this, the mapped vector of each patch is similar to a token, used to represent the local information of the eye detection sub-image. The eye detection sub-features obtained in this application can include information such as eye texture, shape, and edges, which will be converted into numerical feature vectors.

[0135] In step S430 of some embodiments, after obtaining multiple eye detection sub-features, the multiple eye detection sub-features are concatenated according to image position data to obtain the object eye detection features corresponding to the initial eye detection image. In this way, the obtained object eye detection features can accurately represent the eye features of the target object and are better used for eye detection, eye recognition or other related computer vision tasks.

[0136] In step S240 of some embodiments, after obtaining the initial dialogue features, object eye measurement features, and object eye detection features, object information encoding is performed on the initial dialogue features, object eye measurement features, and object eye detection features according to the object information encoder in the feature extraction sub-model to obtain the initial object eye features. This application inputs multiple feature data from different sources into a preset object information encoder, and through the conversion of the object information encoder, these features are transformed into a unified feature representation, namely the initial object eye features. At this point, the initial object eye features integrate information from different sources, have better expressive power and discriminative power, reduce dependence on a single feature, and help improve the generalization ability of the model, thereby improving the accuracy of intraocular lens recommendations.

[0137] For example, such as Figure 5 As shown, the initial data for the target object includes initial dialogue data, initial eye measurement data, and initial eye detection images. The initial dialogue data is segmented into dialogue questions, resulting in multiple dialogue sub-data 510. Text features are extracted from each dialogue sub-data 510, resulting in dialogue text sub-features 511. Multiple dialogue text sub-features 511 are concatenated to obtain initial dialogue features 512. The initial eye measurement data is segmented into measurement data, resulting in multiple eye measurement sub-data 520. Text features are extracted from each eye measurement sub-data 520, resulting in eye measurement text sub-features 521. Multiple eye measurement text sub-features 521 are concatenated to obtain the target eye measurement features 522. The initial eye detection images include corneal topography 530 and fundus optical coherence tomography scans 540. Each initial eye detection image is segmented to obtain a first eye detection sub-image 531 corresponding to the corneal topography map 530, a second eye detection sub-image 541 corresponding to the fundus optical coherence tomography scan 540, and image position data for each eye detection sub-image. Each eye detection sub-image is mapped to a word to obtain eye detection sub-features. Then, multiple eye detection sub-features corresponding to each initial eye detection image are concatenated to obtain object eye detection features 550. After obtaining the initial dialogue features, object eye measurement features, and object eye detection features, these features are input into an object information encoder for object information encoding to obtain the initial object eye features.

[0138] In step S130 of some embodiments, the historical object library is data constructed based on object information of multiple historical objects and the type of intraocular lens used in the surgery. Historical data is used to characterize data associated with historical objects. A historical object refers to a patient who has previously undergone surgery. The historical object data is similar to the initial data described above, except that it exists here as data of historical objects. The historical object data includes historical dialogue data, historical eye measurement data, and historical eye detection images. The historical dialogue data, historical eye measurement data, and historical eye detection images are the same as the initial dialogue data, initial eye measurement data, and initial eye detection images described above, except that the target object is replaced with the historical object, which will not be repeated here. The historical intraocular lens category refers to the type of lens used by the historical object during the surgery.

[0139] In step S140 of some embodiments, to improve the efficiency of intraocular lens (IOL) recommendation, this application can first find historical objects with high similarity based on the object's association information, obtain the IOL category used by the historical object in surgery, and improve the accuracy of IOL recommendation for the target object. Therefore, in the first matching, the historical object matching sub-model of the lens recommendation model is used to match the historical object data based on the comprehensive initial object eye features to determine multiple historical data with high similarity from the historical object database. The historical object data with high similarity is used as the target object data, and the historical IOL category corresponding to the target object data is used as the first predicted IOL category.

[0140] It should be noted that this application can place the historical object database outside the overall model. In this way, the historical object database can be continuously updated as the number of historical objects and corresponding data increases, without the need to retrain the network, thus improving the efficiency of intraocular lens recommendation.

[0141] Please see Figure 6 , Figure 6 This is a specific flowchart of step S140 provided in the embodiments of this application. In some embodiments of this application, step S140 may specifically include, but is not limited to, steps S610 to S640.

[0142] Step S610: Extract object eye features based on historical dialogue data, historical eye measurement data, and historical eye detection images to obtain historical object eye features;

[0143] Step S620: Calculate the similarity between the initial object's eye features and the historical object's eye features to obtain the object's eye similarity value;

[0144] Step S630: Sort the historical data according to the similarity value of the object's eyes to obtain the historical data sequence;

[0145] Step S640: Select data from the historical data sequence according to the preset similarity threshold to obtain the target data, and determine the target object data according to the historical object data corresponding to the target data.

[0146] In steps S610 and S620 of some embodiments, similar to the calculation process of the initial object eye features, the feature extraction sub-model is used to extract object eye features from historical dialogue data, historical eye measurement data, and historical eye detection images to obtain historical object eye features. Historical object eye features refer to the features extracted after integrating multiple types of data and images of historical objects. To find historical objects with high similarity to the target object, similarity is calculated based on the initial object eye features and historical object eye features to obtain an object eye similarity value. The object eye similarity value is a numerical value determined after integrating the dialogue data, eye measurement data, and eye detection images of historical and target objects to characterize the degree of similarity between the historical and target objects.

[0147] It should be noted that methods for calculating the similarity between the eye features of the initial object and the eye features of historical objects include, but are not limited to, Euclidean distance, cosine similarity, and correlation coefficients (such as Pearson correlation coefficient, Spearman correlation coefficient, etc.). In practical applications, the choice of similarity calculation method depends on factors such as specific data characteristics, task requirements, and computational efficiency. Depending on the specific circumstances, a suitable similarity calculation method can be selected to measure the similarity between the eye features of the initial object and the eye features of historical objects.

[0148] In step S630 of some embodiments, after obtaining the object eye similarity value corresponding to each historical data point, the historical data is sorted according to the object eye similarity value. The data sorting method can be ascending or descending order to obtain a historical data sequence.

[0149] In step S640 of some embodiments, the preset similarity threshold refers to a pre-set maximum number of historical data with high similarity to be selected. Regardless of whether the historical data sequence is sorted in descending or ascending order, data selection is performed from the historical data with the highest similarity value of the object's eyes to select the preset similarity threshold of historical data, and these data are used as target data. Furthermore, the historical object data corresponding to each target data is used as the target object data.

[0150] It should be noted that the preset similarity threshold must be less than the number of historical data contained in the historical object library; otherwise, the historical object matching process would be meaningless.

[0151] It should be noted that a matching lens set can be constructed based on multiple first predicted intraocular lens categories determined by the matching.

[0152] In step S150 of some embodiments, the lens recommendation library refers to a structure used to store different categories of artificial lenses. The lens recommendation library stores multiple candidate artificial lens categories and candidate lens parameters (such as material, shape design, fixation method, optical parameters, etc.) corresponding to each candidate artificial lens category.

[0153] It should be noted that the intraocular lens (IOL) market is constantly evolving, and IOL categories are continuously being updated and iterated. This application places the historical object database and the IOL recommendation database outside the model. After a training cycle, updating the historical object database and the IOL recommendation database does not require retraining the model network, thus effectively improving the efficiency of IOL recommendation. Furthermore, updating the historical object database and the IOL recommendation database also improves the accuracy of IOL recommendations.

[0154] In step S160 of some embodiments, after determining the first predicted intraocular lens category, this application may employ a lens category matching sub-model to perform lens category matching on candidate intraocular lens categories based on the initial object eye features and the first predicted intraocular lens category, so as to determine the target intraocular lens recommendation category from the candidate intraocular lens categories.

[0155] Please see Figure 7 , Figure 7 This is a specific flowchart of step S160 provided in the embodiments of this application. In some embodiments of this application, step S160 may specifically include, but is not limited to, steps S710 to S730.

[0156] Step S710: Extract lens attribute features from the first predicted intraocular lens category to obtain the predicted intraocular lens attribute features;

[0157] Step S720: Perform feature fusion on the initial object eye features and the predicted intraocular lens attribute features to obtain the target object eye features;

[0158] Step S730: Match the candidate artificial lens categories based on the eye characteristics of the target object to determine the recommended category of the target artificial lens.

[0159] In step S710 of some embodiments, the predicted intraocular lens attribute features are used to characterize the features obtained after extracting attribute parameters from the first predicted intraocular lens category. The predicted intraocular lens attribute features may include features encompassing all parameters of the first predicted intraocular lens category; alternatively, they may be key features selected from the features of all parameters of the first predicted intraocular lens category. By excluding features with minimal impact on the prediction target, model complexity is reduced and prediction performance is improved; no specific limitation is made here. Therefore, this application can extract lens attribute features for each first predicted intraocular lens category according to a preset lens attribute encoder to obtain the corresponding predicted intraocular lens attribute features.

[0160] In step S720 of some embodiments, after determining the predicted intraocular lens attribute features corresponding to each first predicted intraocular lens category, feature fusion is performed on the initial object eye features and each predicted intraocular lens attribute feature to obtain multiple target object eye features. Therefore, the obtained target object eye features can pass on the unique information of the target object and be fused with the data of the first predicted intraocular lens category with high similarity in the historical object database.

[0161] It should be noted that since the number of the first predicted intraocular lens categories is a preset similarity threshold, the number of the target object's eye features obtained is also a preset similarity threshold.

[0162] In step S730 of some embodiments, after determining multiple target object eye features, lens category matching is performed on candidate artificial lens categories based on the target object eye features to determine the recommended category of target artificial lens from the candidate artificial lens categories.

[0163] Please see Figure 8 , Figure 8 This is a specific flowchart of step S730 provided in the embodiments of this application. In some embodiments of this application, step S730 may specifically include, but is not limited to, steps S810 to S830.

[0164] Step S810: Extract lens attribute features from candidate artificial lens categories to obtain candidate artificial lens attribute features;

[0165] Step S820: Calculate the similarity between the eye features of the target object and the attribute features of the candidate artificial lens to obtain the lens matching value;

[0166] Step S830: Determine the recommended category of the target intraocular lens from the candidate intraocular lens categories based on the lens matching degree value.

[0167] In steps S810 to S830 of some embodiments, lens attribute features are extracted for the candidate lens parameters of each candidate artificial lens category to obtain candidate artificial lens attribute features. Similarity is calculated sequentially for each target object's eye features and each candidate artificial lens attribute feature to obtain multiple lens matching scores. Thus, the candidate artificial lens category corresponding to the largest lens matching score among the multiple lens matching scores can be used as the recommended category for the target artificial lens.

[0168] It should be noted that after calculating the similarity between the eye features of each target object and the attribute features of each candidate artificial lens (IOL), one candidate IOL category may correspond to multiple lens matching values. In this case, the average of the multiple lens matching values ​​corresponding to one candidate IOL category can be calculated to obtain the final target lens matching value corresponding to that candidate IOL category. Thus, the candidate IOL category corresponding to the largest lens matching value among multiple target lens matching values ​​can be used as the recommended category for the target IOL.

[0169] It should be noted that in practical applications, multiple candidate IOL categories can be sorted in descending order based on lens matching scores to obtain a lens category sequence. The first k (k is a positive integer greater than 1) candidate IOL categories in the lens category sequence are then pushed to the doctor to assist the doctor in selecting based on other individual circumstances of the target patient, in order to determine the final recommended category of the target IOL.

[0170] Therefore, in the second matching, the lens category matching sub-model of the lens recommendation model is used. Based on the comprehensive initial eye features of the target, the candidate IOL categories in the lens recommendation library are matched to determine multiple IOL recommendation categories with high similarity from the lens recommendation library. At this point, the IOL recommendation category with high similarity can be used as the target IOL recommendation category, or one of the multiple highly similar IOL recommendation categories can be selected as the target IOL recommendation category according to actual needs; no specific limitation is made here.

[0171] In one embodiment, step S830 may include, but is not limited to, the following steps:

[0172] Obtain candidate evaluation data for candidate lens categories;

[0173] Based on the candidate evaluation data, the lens matching degree value is matched and evaluated to determine the target matching degree value;

[0174] The target intraocular lens (IOL) category is determined from the candidate IOL categories based on the target matching score.

[0175] Understandably, candidate evaluation data refers to data obtained in advance from doctors, intraocular lens manufacturers, or publicly available online lens reviews regarding the evaluation of candidate intraocular lens categories. Candidate evaluation data can include the technical parameters, materials, manufacturer information, and user experiences of the candidate lens category. Obtaining candidate evaluation data helps doctors or target individuals better understand the specifics of the candidate lens category, enabling them to match the evaluation and select a more suitable intraocular lens.

[0176] It is understandable that the target matching value refers to the numerical value determined after fusing the output lens matching value and the candidate evaluation data. This allows for a better determination of the degree of matching between the candidate artificial lens type and the target object's eye.

[0177] Therefore, determining the recommended category of the target intraocular lens (IOL) from the candidate IOL categories based on the target matching score can better assist doctors in selecting a more suitable IOL, thereby improving the surgical outcome and the quality of life of the target patient.

[0178] In one embodiment, the step of performing a matching evaluation on the lens matching degree value based on the candidate evaluation data to determine the target matching degree value may include, but is not limited to, the following steps:

[0179] Feature extraction is performed on the candidate evaluation data to obtain candidate evaluation features;

[0180] The candidate evaluation features are normalized to obtain the candidate evaluation values;

[0181] The target matching value is obtained by weighting the candidate evaluation value and the lens matching value.

[0182] As can be understood, candidate evaluation features refer to the feature forms that characterize candidate evaluation data. In this context, candidate evaluation features can represent the feature forms of key information extracted from candidate evaluation data, effectively reducing unnecessary information and noise, and improving the efficiency and accuracy of subsequent processing.

[0183] Understandably, normalizing the extracted candidate evaluation features can unify the value ranges of different features to the same scale. The candidate evaluation values ​​then represent values ​​mapped to the same scale, such as within the range [0-1]. Therefore, by normalizing features, differences in units and value ranges between different features can be eliminated, which is beneficial for subsequent evaluation and matching calculations, and improves the comparability and interpretability of the data.

[0184] Understandably, by weighting the candidate evaluation values ​​and the lens matching values, the importance of both can be comprehensively considered, resulting in a more comprehensive and objective target matching value, which helps in selecting the most suitable type of intraocular lens.

[0185] like Figure 9 As shown, combined with Figure 5 After obtaining the initial dialogue features, object eye measurement features, and object eye detection features, these features are input into the object information encoder for object information encoding to obtain the initial object eye features. A historical object database 910 is acquired, which stores multiple historical data sets. Object eye features are extracted from these historical data sets to obtain multiple historical object eye features. Similarity is calculated between the initial object eye features and the historical object eye features to obtain object eye similarity values. The historical data is then sorted according to these similarity values ​​to obtain a historical data sequence. Data selection is performed on the historical data sequence based on a preset similarity threshold to obtain target data. The target object data is then determined based on the historical object data corresponding to the target data. The historical intraocular lens (IOL) category corresponding to the target object data is identified as the first predicted IOL category. At this point, a matching lens set 920 can be constructed based on the multiple first predicted IOL categories determined through matching. Lens attribute features are extracted from each first predicted IOL category in the matching lens set 920 using a preset lens attribute encoder to obtain the corresponding predicted IOL attribute features 930. Feature fusion is performed on the initial object's eye features and the predicted intraocular lens (IOL) attribute features 930 to obtain multiple target object eye features 940. An IOL recommendation library 950 is acquired, containing multiple candidate data, each including a candidate IOL category and corresponding candidate IOL parameters. Based on the target object's eye features, IOL category matching is performed on the candidate IOL categories to determine the recommended IOL from the candidate IOL categories. This recommended IOL includes multiple IOL recommendation categories. At this point, the target IOL recommendation category can be determined from the multiple IOL recommendation categories.

[0186] The intraocular lens (IOL) recommendation method provided in this application combines a historical object database and an IOL recommendation database for IOL recommendation. This application can first find historical objects with high similarity based on the object's association information, and obtain the IOL category used by that historical object in surgery, thereby improving the accuracy of IOL recommendations for the target object. Specifically, in the first matching, a historical object matching sub-model of the IOL recommendation model is used to match historical object data based on the comprehensive initial object eye features to determine multiple historical data with high similarity from the historical object database. The historical object data with high similarity is used as the target object data, and the historical IOL category corresponding to the target object data is used as the first predicted IOL category. Each first predicted IOL category is encoded by a lens attribute encoder, and the encoded features can be fused with the comprehensive initial object eye features to pass on the unique information of the target object. In the second matching process, a lens category matching sub-model of the lens recommendation model is used to match candidate IOL categories in the lens recommendation library based on the comprehensive initial object eye features, thereby determining multiple IOL recommendation categories with high similarity from the lens recommendation library. At this point, the IOL recommendation category with high similarity can be used as the target IOL recommendation category, or one can be selected from multiple highly similar IOL recommendation categories as the target IOL recommendation category according to actual needs. Based on the two matching processes, the accuracy of lens matching can be effectively improved. This application places the historical object library and lens recommendation library outside the model. After one training cycle, updating the historical object library and lens recommendation library does not require retraining the model network, thereby effectively improving the efficiency of IOL recommendation. Furthermore, updating the historical object library and lens recommendation library also improves the accuracy of IOL recommendation. Furthermore, when making lens recommendations, this application not only considers the postoperative effects of lenses based on the historical patient database, but also takes into full account various types of information such as the patient's lifestyle habits and predicted consumption based on the initial dialogue data, effectively improving the accuracy of intraocular lens recommendations.

[0187] Please see Figure 10 This application also provides an intraocular lens recommendation device, which can implement the above-described intraocular lens recommendation method. The device includes:

[0188] The data acquisition module 1010 is used to acquire the initial dialogue data, initial eye measurement data and initial eye detection image of the target object;

[0189] The feature extraction module 1020 is used to extract object eye features based on initial dialogue data, initial eye measurement data and initial eye detection images to obtain initial object eye features;

[0190] The object library acquisition module 1030 is used to acquire a preset historical object library. The historical object library includes historical data, which includes historical object data and the corresponding historical intraocular lens categories.

[0191] The object matching module 1040 is used to perform historical object matching on historical object data based on the initial object eye features, determine the target object data, and determine the first predicted intraocular lens category based on the historical intraocular lens category corresponding to the target object data.

[0192] The recommendation library acquisition module 1050 is used to acquire a preset lens recommendation library, which includes candidate artificial lens categories;

[0193] The lens matching module 1060 is used to perform lens category matching on candidate intraocular lenses based on the initial eye features of the object and the first predicted intraocular lens category, and to determine the target intraocular lens recommended category. The target intraocular lens recommended category is used to represent the intraocular lens category recommended for the target object.

[0194] The specific implementation of the intraocular lens recommendation device in this application is basically the same as the specific implementation of the intraocular lens recommendation method described above, and will not be repeated here.

[0195] This application also provides an electronic device, comprising: at least one memory; at least one processor; and at least one computer program; wherein the at least one computer program is stored in the at least one memory, and the at least one processor executes the at least one computer program to implement the above-described recommended method for intraocular lens implantation. This electronic device can be any smart terminal, including tablet computers, in-vehicle computers, etc.

[0196] Please see Figure 11 , Figure 11 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes:

[0197] The processor 1110 can be implemented using a general-purpose central processing unit (CPU), microprocessor, application specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application.

[0198] The memory 1120 can be implemented as a read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory 1120 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 1120 and called and executed by the processor 1110 to execute the recommended method for artificial lens in the embodiments of this application.

[0199] The input / output interface 1130 is used to implement information input and output;

[0200] The communication interface 1140 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0201] Bus 1150 transmits information between various components of the device (e.g., processor 1110, memory 1120, input / output interface 1130, and communication interface 1140);

[0202] The processor 1110, memory 1120, input / output interface 1130 and communication interface 1140 are connected to each other within the device via bus 1150.

[0203] This application also provides a computer-readable storage medium storing a computer program for causing a computer to execute the above-described recommended method for intraocular lenses.

[0204] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0205] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0206] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0207] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; 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 embodiment according to actual needs.

[0208] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0209] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0210] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0211] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0212] The units described above as separate components may or may not be physically separate. 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 units can be selected to achieve the purpose of this embodiment according to actual needs.

[0213] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0214] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0215] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A method for recommending intraocular lenses, characterized in that, The method includes: Acquire initial data of the target object, including initial dialogue data, initial eye measurement data, and initial eye detection images; Based on the initial dialogue data, the initial eye measurement data, and the initial eye detection image, object eye features are extracted to obtain initial object eye features; Obtain a preset historical object library, which includes historical data, including historical object data and the corresponding historical intraocular lens categories; Based on the initial object's eye features, the historical object data is matched with historical objects to determine the target object data, and based on the historical intraocular lens category corresponding to the target object data, the first predicted intraocular lens category is determined. Obtain a preset lens recommendation library, which includes candidate artificial lens categories; Lens attribute features are extracted from the first predicted intraocular lens category to obtain predicted intraocular lens attribute features; feature fusion is performed on the initial object's eye features and the predicted intraocular lens attribute features to obtain the target object's eye features; lens attribute features are extracted from the candidate intraocular lens categories to obtain candidate intraocular lens attribute features; similarity is calculated between the target object's eye features and the candidate intraocular lens attribute features to obtain a lens matching degree value; and a target intraocular lens recommendation category is determined from the candidate intraocular lens categories based on the lens matching degree value, wherein the target intraocular lens recommendation category is used to represent the intraocular lens category recommended for the target object.

2. The method according to claim 1, characterized in that, The step of extracting object eye features based on the initial dialogue data, the initial eye measurement data, and the initial eye detection image to obtain initial object eye features includes: Based on the initial dialogue data, dialogue features are extracted to obtain the initial dialogue features; Based on the initial eye measurement data, measurement data is extracted to obtain the object's eye measurement features; Image feature extraction is performed based on the initial eye detection image to obtain the object's eye detection features; The initial object eye features are obtained by encoding the initial dialogue features, the object eye measurement features, and the object eye detection features using a preset object information encoder.

3. The method according to claim 2, characterized in that, The step of extracting dialogue features based on the initial dialogue data to obtain initial dialogue features includes: The initial dialogue data is segmented into dialogue questions to obtain dialogue sub-data; Text features are extracted from the dialogue sub-data to obtain dialogue text sub-features; The initial dialogue features are obtained by concatenating the sub-features of the dialogue text.

4. The method according to claim 2, characterized in that, The step of extracting image features from the initial eye detection image to obtain the object's eye detection features includes: The initial eye detection image is segmented to obtain an eye detection sub-image and image position data of the eye detection sub-image; Local image encoding is performed on the eye detection sub-image to obtain eye detection sub-features; The eye detection sub-features are spliced ​​together based on the image location data to obtain the object's eye detection features.

5. The method according to claim 1, characterized in that, The historical object data includes historical dialogue data, historical eye measurement data, and historical eye detection images; The step of performing historical object matching on the historical object data based on the initial object's eye features to determine the target object data includes: Based on the historical dialogue data, the historical eye measurement data, and the historical eye detection images, object eye features are extracted to obtain historical object eye features; The similarity between the initial object's eye features and the historical object's eye features is calculated to obtain the object's eye similarity value. The historical data is sorted according to the similarity value of the object's eyes to obtain a historical data sequence; The historical data sequence is selected based on a preset similarity threshold to obtain target data, and the target object data is determined based on the historical object data corresponding to the target data.

6. A device for recommending intraocular lenses, characterized in that, The device includes: The data acquisition module is used to acquire initial data of the target object, including initial dialogue data, initial eye measurement data, and initial eye detection images; The feature extraction module is used to extract object eye features based on the initial dialogue data, the initial eye measurement data, and the initial eye detection image to obtain initial object eye features; The object library acquisition module is used to acquire a preset historical object library, which includes historical data, including historical object data and the corresponding historical intraocular lens categories; The object matching module is used to perform historical object matching on the historical object data based on the initial object eye features to determine the target object data, and to determine the first predicted intraocular lens category based on the historical intraocular lens category corresponding to the target object data. The recommendation library acquisition module is used to acquire a preset lens recommendation library, which includes candidate artificial lens categories; The lens matching module is used to extract lens attribute features from the first predicted intraocular lens category to obtain predicted intraocular lens attribute features; perform feature fusion on the initial object's eye features and the predicted intraocular lens attribute features to obtain target object's eye features; extract lens attribute features from the candidate intraocular lens categories to obtain candidate intraocular lens attribute features; calculate the similarity between the target object's eye features and the candidate intraocular lens attribute features to obtain a lens matching degree value; and determine a target intraocular lens recommendation category from the candidate intraocular lens categories based on the lens matching degree value, wherein the target intraocular lens recommendation category is used to represent the intraocular lens category recommended for the target object.

7. An electronic device, characterized in that, include: At least one memory; At least one processor; At least one computer program; The at least one computer program is stored in the at least one memory, and the at least one processor executes the at least one computer program to perform: The method as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that is used to cause the computer to execute: The method as described in any one of claims 1 to 5.

Images