Multi-focus contact lens prediction system based on machine learning

Abstract

A method for automatically determining multifocal lens parameters associated with a contact lens and a contact lens for correcting ametropia of a patient's eye wherein a personalized multifocal contact lens for treating presbyopia takes into account various related clinical patient measurements including patient clinical data and lifestyle data, to determine which lens is likely to provide personalized lens parameters for new patients. Historical patient clinical data and lifestyle data are used to train the system to predict which multi-focal lenses (i.e., lenses including lens designs and having lens geometries and powers) would result in optimal lens-to-eye fit and optimal visual function for each eye of presbyopia patients.

Description

Technical Field

[0001] The disclosed technology relates generally to contact lenses, and more particularly, some embodiments relate to multifocal contact lenses and systems for determining the design and parameters of multifocal contact lenses. Background Art

[0002] A normal eye produces clear images by bending (refracting) light rays to focus them on the retina. Refractive errors are a common cause of impaired vision. Refractive errors occur when the eye's optical components are unable to properly focus light rays from objects 20 feet or more away onto the retina, the light-sensitive tissue layer at the back of the eye. Light rays from objects 20 feet or more away from the eye are called parallel rays.

[0003] Even without refractive error, the eye must perform additional focusing on objects within 20 feet. The closer the object is to the eye, the greater the focusing power required. This focusing ability of the eye is called accommodation. Accommodation refers to the eye's ability to increase the optical power of the lens to focus close objects on the retina. Presbyopia is a physiological inability of the eye to accommodate itself that occurs with aging, resulting in a gradual decrease in the ability to focus clearly on close objects. The most significant decrease in the eye's ability to accommodate occurs between the ages of 20 and 60. By age 45, presbyopia begins to interfere with the ability to comfortably read small print at a normal reading distance without lens correction.

[0004] As the baby boomer generation approaches the age of presbyopia, the demand for corrective lenses for presbyopia is increasing dramatically. Framed eyeglasses, such as reading glasses, bifocals, or progressive lenses, are the most common forms of presbyopia correction. Contact lenses for presbyopia have yet to gain widespread market acceptance. Rather than using bifocal or multifocal contact lenses, some eye care practitioners employ so-called monovision correction, which involves using a contact lens for distance vision in one eye and a contact lens for near vision in the other. Most multifocal contact lenses are mass-produced and not custom-made for each patient's eye. Specifically, most manufactured multifocal contact lenses have a single zone diameter and do not include features that place the optical center of the lens on the visual axis of the wearer's eye. Currently, multifocal contact lenses that are centered on the visual axis and guided by pupil size are rarely used for presbyopia correction. These lenses typically have a first zone with optical power for correcting the patient's vision when focusing on near objects, and a second zone with optical power for correcting vision when focusing on distant objects. The optical zone must be moved to align with the visual axis of the eye, and the diameters of the near and distance zones must be adjusted based on pupil size measurements.

[0005] Because the location and size of these zones are typically determined based on demographic or physiological prevalence, these lenses lack the level of customization offered by custom eyeglasses, which are precisely fitted for each eye and patient. Although multifocal contact lenses can take into account certain clinical measurements, such as measurements of pupil size, pupil reactivity, and pupil movements, these lenses fail to incorporate other clinical measurements necessary to create personalized multifocal contact lenses for presbyopia correction.

[0006] Therefore, there is a need for efficient fitting and manufacturing methods that take into account various clinical factors, customize multifocal contact lenses without increasing costs and while using existing manufacturing methods. There is a need for machine learning systems that use input data and algorithms to predict recommended lens shape factors and lens parameters, and utilize the resulting data to further improve the system's predictive capabilities. Summary of the Invention

[0007] The present disclosure provides a system and method for generating personalized multifocal contact lenses, wherein the personalized multifocal contact lenses have an optimal lens-eye fitting relationship and optimal visual function for each eye of a presbyopic patient, so as to solve the pain points in the industry.

[0008] In one aspect of the present disclosure, a computer-implemented method may include receiving a plurality of data, the plurality of data including one or more patient eye characteristics, patient physical characteristics, patient activity data, and patient demographic data. The method may also include determining a classification value and a confidence level for a category by applying a machine learning model, wherein the plurality of data serves as input for each category of the plurality of data. In some embodiments, the method may include determining a weighted multifocal lens parameter by combining the classification value and the confidence level for each category of the plurality of data. Finally, the method may include providing information associated with determining the weighted multifocal lens parameter. In some embodiments, the determination of the weighted multifocal lens parameter may be displayed via a graphical user interface (GUI) that may include a display element.

[0009] In some embodiments, the multifocal lens parameters may include lens design and lens geometry. For example, the lens design may include a refractive design, a translational design, a diffractive design, a diffractive-refractive design, a pinhole aperture design, a spiral optical design, and / or an intelligent eye accommodation design. Similarly, the lens geometry may include the area of ​​a first zone of the contact lens, the distance and angular position from the first zone to the geometric center of the second zone, the overall diameter, the base curve radius, the posterior surface peripheral sagittal depth asymmetry, the anterior surface anti-rotation surface features, the thickness profile, and / or surface modification measurements.

[0010] In some embodiments, the multifocal lens parameters may include the displacement or decentering of a zone of the contact lens, the diameter of the zone, the power of the zone, and the power profile of the zone.

[0011] Implementations of the disclosed technology may include one or more of the following features. For example, the multiple data used as input by the machine learning model may include patient eye characteristics, which may include one or more of subjective refraction measurements, computerized autorefractometry measurements, ocular aberration measurements, accommodative amplitude measurements, depth of field measurements with reading addition, corneal topography, ocular surface tomography, keratometric measurements, eyelid position measurements, palpebral fissure size, tear film break-up time, pupil size, and kappa angle measurements. Furthermore, the multiple data used as input by the machine learning model may include patient physical characteristics, such as the patient's height, arm length, age, sex assigned at birth, and / or race. Finally, the multiple data used as input by the machine learning model may include patient activity data, which may include responses to interview questions used to determine the patient's eye use. For example, the patient may indicate the number of hours they typically spend on work and leisure activities, and the lighting conditions associated with these work and leisure activities. Furthermore, the patient may provide viewing distances and working distances, as well as the height at which objects are viewed during work and leisure activities. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The technology disclosed herein is described in detail according to one or more different embodiments with reference to the following figures. The figures are provided for illustrative purposes only and depict only typical or exemplary embodiments of the disclosed technology. These figures are provided to facilitate the reader's understanding of the disclosed technology and should not be construed as limiting its breadth, scope, or applicability. It should be noted that for clarity and ease of illustration, these figures are not necessarily drawn to scale.

[0013] Figure 1 An embodiment of a multifocal contact lens parameter prediction system according to implementations of the present disclosure is shown.

[0014] Figure 2A A block diagram illustrating the functions implemented by a multifocal contact lens parameter prediction system according to some embodiments of the present disclosure.

[0015] Figure 2B Some embodiments of the present disclosure are shown Figure 2A Block diagram of the multifocal contact lens parameter prediction system shown.

[0016] Figure 3 is a flow chart illustrating an overview process of multiple focus zone parameters according to some embodiments of the present disclosure.

[0017] Figure 4 A block diagram depicts an example computer system in which embodiments described herein may be implemented. DETAILED DESCRIPTION

[0018] Described herein are systems and methods for improving lens customization by providing an improved method for determining parameters of multiple focal zones using a limited set of input data. Details of some example embodiments of the disclosed systems and methods are set forth in the following description. Other features, objects, and advantages of the present disclosure will become apparent to those skilled in the art upon examination of the following description, drawings, examples, and claims. It is intended that all such additional systems, methods, features, and advantages be included within this specification, be within the scope of this disclosure, and be protected by the accompanying claims.

[0019] As described above, by using a system that includes input data and clinical outcome measures, the shape and parameters of multifocal contact lenses can be predicted, which provides patients with personalized and optimized correction for their refractive error and presbyopia without having to rely on time-consuming examinations by practitioners and the resulting variable success rates of fitting glasses.

[0020] The contact lenses envisioned by the disclosed invention can be any lens that contacts ocular tissue, including soft, rigid, and hybrid contact lenses, intracorneal lenses, and intraocular lenses. The contact lenses have an anterior surface, a posterior surface, and a medium therebetween. The contact lenses can have fully or partially encapsulated components including microelectromechanical systems, electro-optical systems, electrochromic systems, or passive fully or partially encapsulated lenses, apertures, or filters.

[0021] The most commonly fitted multifocal contact lens is a central near vision lens, which has a central power zone for reading and intermediate distance tasks, surrounded by a second annular power zone that corrects the eye's refractive error to provide clear distance vision. The two or more zones of different power are usually made on the front surface and are created by setting different radii of curvature. The two or more zones of different power can also be located on the back surface or in the medium between the front and back surfaces by varying the refractive index.

[0022] In some embodiments, the multifocal optical device may be refractive, while in other embodiments, the multifocal optical device may be diffractive, helical, or a combination of refractive, diffractive, helical, or electro-optical, electrochromic, or aperture optical devices. The disclosed invention may be used to apply machine learning to predict a class of optical configuration or to select at least one of a zone location, a zone power, or a zone power profile for a multifocal contact lens.

[0023] Multifocal contact lenses can include different zones designated for distance vision, near vision, and sometimes intermediate vision. For example, a lens can include a primary viewing zone located in the center of the lens that is surrounded by concentric rings having near and distance vision powers. The central viewing zone can be used for viewing distant objects and can be referred to as a central distance vision design. In contrast, some multifocal contact lenses can include a central viewing zone for viewing close objects, referred to as a central near vision design, and at least one zone surrounding the central viewing zone has optical power for correcting distance vision. In some cases, the central distance vision design can be used for the patient's dominant eye and the central near vision design can be used for the non-dominant eye, or less often, the opposite configuration can be used.

[0024] One aspect of the disclosed embodiment features a multifocal contact lens comprising (1) a first zone centered on the visual axis of a patient's eye and having a first optical power and a first optical power profile; and (2) at least one outer zone peripheral to the first zone and having a second optical power and a second optical power profile. Notably, the second optical power of the outer zone can be different from the first optical power of the first zone. The outer zone can include a distance and an angular position from the first zone to a geometric center of the outer zone, the distance and angular position being selected to place at least the first zone on the visual axis of the user's eye and to position the second zone concentrically with the first zone.

[0025] Furthermore, both the central zone and the peripheral zone may include at least one of a spherical lens power, an aspherical lens power, or a toric lens power. For example, the aspherical lens power may include a conic constant for defining a power profile or a linear distortion profile, wavefront-guided high-order aberration correction, spiral optics, or a diffractive lens power.

[0026] In some embodiments, the geometric center of the first zone can coincide with the geometric center of the peripheral zone, and the first and outer (second) zones can be displaced by a certain distance and angular position relative to a third zone peripheral to the outer (second) zone. In some embodiments, the third peripheral zone is not considered an optical path for vision correction and can be used to create a lens shape optimized for thickness and / or comfort and / or orientation and translation stability and / or enhanced anti-deformation.

[0027] In some embodiments, the multifocal lens can be rotationally stabilized to allow at least one of: different optical powers in two or more semi-meridians; different higher-order aberrations in two or more semi-meridians, or two or more linear distortion forms in two or more semi-meridians.

[0028] The lens design of a multifocal contact lens and the positions and optical powers of the central and peripheral zones can vary according to the individual characteristics of the patient. Therefore, producing a personalized multifocal contact lens for a patient's eye requires: (i) obtaining the patient's (a) clinical information to identify individual eye characteristics; and (b) lifestyle information to identify the patient's physical characteristics and usage characteristics, and (ii) determining the lens design and geometry of the lens and the optical powers and optical power distribution of the central and peripheral zones based on the obtained clinical and lifestyle information.

[0029] In one embodiment of the present invention, a method for obtaining clinical information about individual eye characteristics of a patient may include collecting clinical data, wherein the clinical data includes one or more of subjective refraction, computerized autorefraction, eye aberrations, amplitude of eye accommodation, depth of field under reading addition, corneal topography, ocular surface tomography, corneal curvature, eyelid position, palpebral fissure size, tear film breakup time, photopic pupil size, intermediate vision pupil size and scotopic pupil size, Kappa angle and / or the angle formed by the corneal vertex normal and the pupil center, deviation between the geometric center of the predicted contact lens and the geometric center of the pupil when worn, aberrations when wearing the predicted contact lens, and sphero-cylindrical refraction when wearing the predicted contact lens.

[0030] Similarly, methods for obtaining patient lifestyle information specifying patient physical characteristics and usage characteristics may include: (1) collecting demographic or anthropomorphic information, including the patient's height, age, sex assigned at birth, sitting height, and habitual head tilt angle and viewing angle; and (2) administering a lifestyle questionnaire, quality of life questionnaire, or structured interview to determine the patient's eye and vision use in work and leisure activities and the number of hours habitually spent in the corresponding activities, the lighting conditions for the corresponding activities, the viewing distance and working distance for the corresponding activities, and the viewing height or viewing angle of objects relative to the primary eye position in the postural position for the corresponding activities.

[0031] Finally, a method for defining the geometry and power of a lens for a central zone and a peripheral zone may include: (1) selecting a first zone having an optical center corresponding to the visual axis, vertex normal, or pupil center of a patient's eye, (2) selecting a power area of ​​the first zone, (3) selecting a power distribution of the first zone, (4) selecting a distance and angular position of the first zone relative to the geometric center of an outer peripheral zone, and (5) selecting the power of an outer zone peripheral to the first zone.

[0032] Therefore, when selecting personalized multifocal contact lens parameters for patients with presbyopia, various clinical measurements and other patient information should be considered. For example, differences in patient ocular anatomy, as well as patient height and preferred activities, may influence the choice of lens design and position, optical zone power, and power distribution.

[0033] The area and power distribution of the first zone and the area and power distribution of the outer zone can be determined by anthropomorphic and clinical measurement inputs and based on patient-reported lifestyle information. As described above, patient-reported lifestyle information can be obtained using a questionnaire or a structured interview.

[0034] In some embodiments, information related to the intended wearer's lifestyle can be used to predict lens parameters. For example, a patient who spends more time outdoors, such as participating in sports activities or driving a motor vehicle, may require a smaller central near multifocal zone. In contrast, a patient who spends most of their waking hours indoors and performs tasks within one meter, particularly within arm's reach, may require a larger central near multifocal zone and will tolerate some reduction in distance vision correction or clarity from contact lenses.

[0035] In some embodiments, the questionnaire can be designed to identify work and leisure activities that may be associated with the need for clearer distance vision or clearer near vision, as well as to identify the range of focal lengths required to complete near and intermediate vision tasks. In some embodiments, these lifestyle factors are quantified and used as input to an algorithm to determine the area and power profile of at least one zone of the lens.

[0036] Thus, the system integrates clinical outcome measurements and follow-up questionnaire feedback to modify the lens design and its specific parameters to generate a second set of lens parameter plans. For example, clinical outcome measurements of distance visual acuity, intermediate visual acuity, and near visual acuity in one and / or both eyes; near and intermediate distance ranges for clear vision; and distance and / or near vision on-lens refraction can be collected and used as input to modify contact lens parameters.

[0037] In some embodiments, feedback to subsequent questionnaires or structured interviews is quantified and used as input to the algorithm to determine the area and power distribution of at least one zone of the lens. In some embodiments, the questions can be taken from a validated quality of life questionnaire for visual function. The questions can be for subjective assessments of overall visual acuity, reading visual acuity, driving visual acuity, night visual acuity, counter work visual acuity, and / or questions designed for assessing glare, halo, ghosting, or diplopia. In some embodiments, questions about visual acuity in dim light and visual acuity in bright light can be included. For example, questions from the National Eye Institute Visual Function Questionnaire-25 (VFQ-25) or other questionnaires can be used.

[0038] Increased frequency of reported difficulty with visual tasks can be used to predict which contact lens parameters may need to be changed. For example, reports of difficulty reading street signs or store names can be used to predict a decrease in the diameter of the central near vision zone or an increase in the diameter of the central distance vision zone. In some embodiments, assessment results obtained through questionnaires or structured interviews can be combined with clinical measurements. For example, on-lens refraction measurements during wear can be combined with questionnaire feedback to determine whether vision difficulties are due to residual refractive error or to a reduced modulation transfer function caused by an overly large or undersized optic zone diameter.

[0039] Those skilled in the art of vision correction will appreciate that each outcome assessment question may be associated with at least one of poor lens centering, polar misalignment, instability, power error, power profile error, incorrect diameter of the central near zone or central distance zone, lens deformation, or insufficient surface wetting. In some embodiments, an algorithm may be used to combine clinical outcome data with questionnaire responses to predict changes in at least one parameter of a contact lens used to correct presbyopia.

[0040] All of these clinical measurements and results are obtained through a variety of clinical tests performed by qualified eye care professionals using a variety of clinical devices, as well as questionnaires or structured interviews. Because each practitioner has varying levels of experience and training, lens designs, lens geometries, and focal powers determined based on the resulting clinical and lifestyle information can vary. This undesirable variability is further exacerbated by the fact that each clinical patient measurement is associated with its own coefficient of variation, standard deviation, or observer error. Cost control and efficient healthcare delivery require high first-time prescription success rates. For these reasons, currently available contact lens treatments for presbyopia do not offer an efficient and cost-effective solution. Therefore, multifocal contact lenses offer the potential to become a lifestyle modification approach that corrects presbyopia using customized lens parameters determined by a machine learning system, significantly reducing prescription errors and optimizing secondary fittings, ensuring time-saving and cost-effective care delivery.

[0041] Machine Learning Overview

[0042] Machine learning is a subset of artificial intelligence that involves algorithms trained to learn patterns and make predictions or decisions without being explicitly programmed. It involves designing and training models to automatically extract patterns, relationships, or rules from data to perform specific tasks. Machine learning encompasses various techniques, such as supervised learning, unsupervised learning, reinforcement learning, and deep learning. The goal of machine learning is to build systems that can improve their performance or behavior through experience or data. Machine learning technology has been applied in a variety of fields, including image and object recognition, natural language processing, medical care and diagnostics, and financial forecasting, and continues to expand rapidly across industries, disrupting business operations and improving efficiency and decision-making processes.

[0043] The following are example steps taken in a machine learning application. Data Collection: The first step is to gather relevant data representing the problem or task at hand. This data is called "input data" and can include examples, observations, or measurements that are already labeled or have known outcomes. Simply put, input data refers to the information provided to a machine learning algorithm for training or prediction. Input data consists of features or attributes that describe the characteristics or properties of the problem at hand. The type and format of input data can vary depending on the specific problem and the nature of the data. Data Preprocessing: Raw data often requires preprocessing and preparation before it can be effectively used. This involves tasks such as cleaning the data, handling missing values, standardizing or scaling features, and splitting the data into training and test sets. Model Selection: Choosing the right machine learning model for the task is crucial. A model is a mathematical representation of the problem, and different models have different capabilities and characteristics. Examples of popular machine learning models include decision trees, support vector machines, neural networks, and random forests. Training the Model: During the training phase, the selected model is fed labeled data to learn (supervised learning). The model analyzes the input data, identifies patterns, and adjusts its internal parameters or weights to minimize the difference between its predicted output and the actual labeled data. Evaluation: Once a model is trained, it needs to be evaluated to gauge its performance. Evaluation is typically done using an independent dataset, called a test set, that the model has not yet seen during training. Performance metrics such as accuracy, precision, recall, or F1 score measure how well the model generalizes to new, unknown data. Model Optimization: If the model's performance is unsatisfactory, adjustments can be made to improve its accuracy. This may involve fine-tuning the model's hyperparameters, selecting different features, or using more advanced techniques (e.g., ensemble methods). Prediction or Decision Making: Once a model has been trained and validated, it can be used to make predictions or decisions on new, unknown input data. The model takes the input data and applies the learned patterns to produce an output, or prediction. Iteration and Improvement: Machine learning is an iterative process. If new data becomes available or the model's performance needs improvement, the process can be repeated by retraining the model with the updated data or making any necessary adjustments.

[0044] Output

[0045] Embodiments of the disclosed technology provide a machine learning algorithm for determining a manufactured multifocal lens that provides optimal eye-eye fit and visual function for a patient and that has a lens design and lens geometry, optical power, and optical power profile. The multifocal lens can be determined by using patient demographic, anthropometric, and psychometric information, clinical information (including multifocal lens fitting), and patient lifestyle information as inputs to train the machine learning algorithm. Furthermore, the patient's blur tolerance can be determined and used in the process of determining the multifocal lens. In some embodiments, the machine learning algorithm can be used to generate recommendations for existing manufactured multifocal lenses that provide optimal eye-eye fit and visual function for the patient. Similarly, blur tolerance can be a factor used in the determination and recommendation generation process. In other embodiments, the machine learning algorithm can be used to generate instructions (e.g., a cutting file) for turning a multifocal lens that provides optimal eye-eye fit and visual function for the patient. Furthermore, the machine learning algorithm can be used to identify existing molds for manufacturing multifocal lenses that provide optimal eye-eye fit and visual function for the patient.

[0046] advantage

[0047] By using machine learning algorithms to determine multifocal lenses, each patient is ensured to receive lenses that provide the optimal eye fit and visual function, regardless of the practitioner used to collect and interpret the data. Thus, when lens design and geometry, as well as the focal power and power profile in the central and peripheral zones of the lens, are determined based on the resulting clinical and lifestyle information, the described embodiments ensure that practitioner variability is minimized, thereby providing a customized multifocal contact lens for each eye of a patient with presbyopia. Furthermore, this result is expected to be achieved by reducing the number of clinical measurements and lifestyle information required as input to the machine learning.

[0048] Embodiments of the disclosed technology provide a machine learning algorithm for determining personalized lens parameters for a specific patient without requiring access to all clinical measurements and / or lifestyle information. Specifically, the machine learning algorithm is trained using historical clinical data collected from clinical observations of patients. For example, these patients may suffer from presbyopia; however, this method for determining personalized lens parameters can provide a continuous and uniform refractive error correction (service). This clinical data includes previously obtained clinical measurements, such as but not limited to demographic information including age and gender; physical characteristics including patient height and patient arm length, and other measurements; activities including computer work and driving, and other lifestyle activities; ocular characteristics including pupil size and prescription correction power, and other clinical measurements. Next, the machine learning algorithm analyzes this data previously provided by the multifocal algorithm, which is used to determine multifocal lens parameters based on relevant previous patient historical characteristics for a specific multifocal lens design. In particular, this method for determining personalized lens fit can achieve a continuous and uniform output for patients with different refractive errors, demographic characteristics, anthropomorphic measurements, and lifestyle vision needs.

[0049] Input Data for Training — Overview

[0050] In some embodiments, the patient history information may include (1) clinical data, including subjective refraction, computerized autorefraction, eye aberrations, accommodative amplitude, depth of field and range of clear vision under reading addition, corneal topography, ocular surface tomography, corneal curvature, eyelid position, palpebral fissure size, tear film breakup time, photopic pupil size, intermediate vision pupil size and scotopic pupil size, Kappa angle, deviation between the geometric center of the predicted contact lens and the geometric center of the pupil, aberrations when wearing the predicted contact lens, and sphero-cylindrical refraction when wearing the predicted contact lens; and (2) patient lifestyle information, including patient height, age, sex assigned at birth and sitting height, and habitual head tilt angle, and responses to patient questionnaires including responses to lifestyle questionnaires, quality of life questionnaires, or structured patient interviews to obtain information about the patient's use of eyes and vision in work and leisure activities and the number of hours habitually spent in the corresponding activities, lighting conditions for the corresponding activities, viewing distance and working distance for the corresponding activities, and viewing height relative to the primary eye position in the posture for the corresponding activities. In other embodiments, a machine learning algorithm can be trained using historical care product regimens, wearing schedules, lens removal methods, and patterns during lens use. This information can be obtained from patient questionnaires, interviews, or other sources. Additionally, input data can include prior multifocal algorithms that describe known relationships between prior patient history characteristics (e.g., age, height, measured reading distance, viewing angle) and corresponding visual function for a specific multifocal lens design (e.g., central near vision or central distance vision) and multifocal contact lens parameters.

[0051] Our use of machine learning – an overview

[0052] Next, as explained in detail herein, the machine learning algorithm model analyzes the relationship between prior multifocal algorithms and multifocal lens parameters based on prior patient history characteristics of the specific multifocal lens design and relevant lens geometry and power. In a preferred embodiment, the first step in algorithm development is to select factors that predict design parameters. Multiple factors may contribute to the prediction of one or more design parameters. Rather than using multiple factors to make the same prediction, the system can identify a single factor that is reliable or effective for predicting one or more parameters. For example, a Donder table can be used that uses only age as a predictor of additional power requirements. Additional power is a spherical plus power superimposed on the fully corrected distance vision diopter to assist presbyopic patients in near vision. Alternatively, a combination of age and height, or age, height, sex assigned at birth, and race may be more appropriate for the initial algorithm to select additional power.

[0053] In some embodiments, an algorithm for selecting the add power can be initially used based on age, height, gender, and race assigned at birth. Lifestyle factors such as time spent indoors versus outdoors, and time spent on central near versus distance tasks can be applied to the algorithm to adjust the add power selection. In some embodiments, pupil size measurements can be incorporated to control the output of the add power zone diameter and the add power profile.

[0054] In some embodiments, the output includes the type or category of multifocal lens design. These categories may include refractive designs, including central near vision, central far vision, translational designs, diffractive designs, diffractive-refractive designs, pinhole aperture designs, spiral optics, and intelligent eye adjustment designs, among others. After the category of the multifocal contact lens is selected, the parameters of the lens are selected. The algorithm uses input data to determine the size of the areas of the contact lens, the power distribution of each area, and the relative positions of the areas within the contact lens. For example, the input may include an image captured of the predicted position of the geometric center of the lens relative to the center of the pupil of the lower eye, the vertex normal of the image of the lower eye, or the visual axis of the lower eye when worn. The data may be used as input, and the algorithm may calculate the amount of optical device displacement required to place the optical device on the pupil center, vertex normal, or visual axis of the corresponding eye.

[0055] In some embodiments, while maintaining the same or similar overall diameter and posterior surface geometry of the predicted lens, including sagittal depth profile and rotational symmetry or asymmetry, the input data can incorporate orientation measurements of the predicted lens. The design algorithm can use these translational and rotational orientation measurements to calculate the displacement of the optics on one or more surfaces of the personalized multifocal lens.

[0056] In some embodiments, the position of the eyelids relative to the eye can be measured. These measurements can include at least one of the position of the upper eyelid relative to the upper limbus, the position of the lower eyelid relative to the lower limbus, the distance between the upper and lower eyelids in primary eye position, the anatomical shape of the underside of the upper and / or lower eyelids, eyelid tension, and the distribution of eyelid thickness from the nasal to the temporal side. An algorithm can use these features to determine features of an anti-rotation design to ensure rotational stability of the lens, improve lens centering, and reduce lens deformation.

[0057] In another embodiment, the first element of the algorithm is to determine the multifocal design category. For example, multifocal design categories may include central near, central distance, alternating rings for distance and near, extended depth of focus, fully aspheric, helical, diffractive, diffractive-refractive, translational, single vision, or modified single vision. The algorithm can be limited to one, two, or more categories, or can include all categories. The following example limits the algorithm output to only central distance or central near design categories.

[0058] Multifocal design category can be determined by combining measured pupil size with medication use known to cause pupil dilation or constriction. A central distance design is appropriate for eyes with larger pupils. For example, a central distance design might be selected for eyes with intermediate vision pupils larger than 3.5 mm and for patients taking medications that cause pupil dilation. Conversely, a central near design might be selected for eyes with pupils smaller than 3.5 mm and for patients taking medications that cause pupil constriction.

[0059] Lifestyle and vision needs history can also be considered when determining the type of multifocal design. A central distance multifocal design may be suitable for patients who have a high need for clear distance vision. A central near design, on the other hand, may be suitable for patients who have a higher tolerance for blurriness, spend a greater proportion of their time indoors, and spend more time on near tasks than distance tasks.

[0060] The viewing angle for near tasks is a strong predictor of multifocal classification. The larger the viewing angle for near tasks and the higher the proportion of the eye in use at this viewing angle, the greater the indication for a central near design. For example, by designating angles above the line of sight directly ahead as positive and angles below the line of sight directly ahead as negative, angles from +40 degrees to -20 degrees indicate suitable lens design for central near. Similarly, angles from -20 degrees to -40 degrees indicate suitable lens design for central distance.

[0061] Next, the algorithm can proceed to select the distance power of the lens. In some embodiments, the initial input can be a non-cycloplegic subjective refraction or a computerized autorefraction. This value can be adjusted based on the type of refractive error. The algorithm can adjust the distance power of the lens by +0.125 diopters for hyperopic refractive errors and -0.125 diopters for myopic refractive errors. In some embodiments, the algorithm can adjust the distance power of the lens based on identified lifestyle factors. For example, patients who frequently engage in outdoor activities, night driving, sports, and viewing can have the distance power of the lens adjusted by -0.25 diopters. For patients who report spending more than 80% of their waking hours indoors, the algorithm can adjust the distance power by +0.125 diopters; and for patients who report spending 95% or more of their waking hours indoors, the power will be adjusted by +0.25 diopters.

[0062] In some embodiments, the algorithm can use pupil size measurements to adjust the distance lens power. For example, if the intermediate or average pupil size is greater than 3mm, the algorithm will adjust the distance lens power by -0.125 power. If the pupil size is less than 2.5mm, the algorithm will adjust the distance lens power by +0.125 power. Optionally, when the intermediate or average pupil size is greater than 3mm, the algorithm can add a conic constant to produce negative spherical aberration.

[0063] In some embodiments, the first input for the near zone power can be based on a biometric average of eye accommodation amplitudes derived from age-based population studies, or the input can be a uniform monocular or binocular cross-cylinder addition measured at a distance of, for example, 40 cm. The algorithm can make adjustments based on the direction of the refractive error. In some embodiments, the near zone power can be adjusted by -0.125 diopters for myopia and +0.25 diopters for cases where the measured accommodation amplitude differs significantly from that of a similarly aged population.

[0064] In some embodiments, the near zone can be adjusted by +0.125 diopters for patients with natal male sex. In some embodiments, the algorithm can adjust the near zone diopters based on the patient's height input. It is well known that taller individuals work at longer distances for near and intermediate distance tasks, while shorter individuals work at shorter distances for near tasks. In some embodiments, the near zone can be adjusted by -0.25 diopters for patients 72 inches or taller, and by +0.25 diopters for patients 62 inches or shorter.

[0065] In some embodiments, the near zone power can be adjusted based on the patient's measured habitual reading distance. For example, for patients with a reading distance of 33 cm or less, the near zone power can be adjusted by +0.25 power, and for patients with a reading distance of 50 cm or more, the near zone power can be adjusted by -0.25 power.

[0066] The amplitude of accommodation has been measured to vary with the viewing angle of the eye, with the largest amplitude of accommodation measured when gazing downward at 40 degrees. In some embodiments, the optical power of the near zone can be adjusted for the patient's habitual viewing angle when performing close eye tasks. The viewing angle is measured from the straight forward line of sight, with positive angles being in the direction of upward gaze and negative angles being in the direction of downward gaze. The algorithm can adjust the optical power of the near zone by +0.125 optical power for viewing angles of 0 to +20 degrees, -0.125 optical power for viewing angles of -20 to -40 degrees, and -0.25 optical power for habitual viewing angles greater than 40 degrees.

[0067] In some embodiments, the algorithm for predicting near zone power may include adjustments for one or more ethnic and racial categories of the patient. For example, Chinese patients have been measured to have a smaller amplitude of accommodation than presbyopic patients of similar age. Therefore, for the eyes of Chinese patients, the algorithm may adjust the near zone power by +0.125 diopter. It is expected that future studies of populations with different ethnic genotypes may reveal differences in the amplitude of accommodation that occur with age. In some embodiments, data from population studies can be used to modify the algorithm to calculate parameters such as near zone power for multifocal contact lenses.

[0068] Central near addition diameter and optical power distribution

[0069] In some embodiments, if the algorithm indicates that a central near multifocal design is used, the next step is to determine the central near add diameter, and if the algorithm indicates that a central distance multifocal design is used, the next step is to determine the central distance diameter. In both cases, the primary input can be pupil size and / or pupil reactivity. The algorithm can use a percentage of the intermediate vision pupil size, an equation that includes the intermediate vision pupil size, a function of the average of the scotopic and photopic pupil sizes, or an equation that factors the intermediate vision pupil size with the range of scotopic to photopic pupil size measurements. In one embodiment, the rule that the algorithm uses to determine the central near add diameter can include calculating at 60% of the intermediate vision pupil size. For example, if the intermediate vision pupil measures 3.3 mm, the central near add diameter is equal to 1.98 mm, which is calculated based on 60% of a 3.3 mm intermediate vision pupil.

[0070] In another embodiment, the algorithm used to determine the central add diameter can be 60% of the average of the scotopic and photopic pupil size measurements. For example, if the scotopic pupil is 4.2 mm and the photopic pupil is 2.9 mm, the average of the two is 3.55; then, by using 60% of the average of 3.55 to calculate the central near add diameter, the central near add diameter is 2.13 mm. In another embodiment, the algorithm can use the larger of the two values ​​or use another mathematical equation that combines the reactivity with or without combining the reactivity with a single measurement of pupil size.

[0071] In some embodiments, a central near add power profile can be determined based on an algorithmically predicted central near add power and an algorithmically predicted central near add diameter. In one embodiment, an algorithmic example includes calculating a conic constant that generates a local slope of the central near add zone to match the local slope of the distance zone at a chord diameter of the central near add diameter. Embodiments can include providing a vertex radius of curvature for the near add that produces the calculated near add power, and then calculating a conic constant that increases the radius of curvature from the vertex to the chord diameter of the central add diameter.

[0072] In some embodiments, the vertex power can be a multiple of the calculated central near add power. For example, the vertex power can be 1.5 times the calculated central near add power. In this embodiment, in order to match the local slope of the distance zone at the chord diameter of the central near add diameter, the calculated cone constant is greater than the cone constant calculated when the vertex power is equal to the calculated near add power. Specifically, when the add diameter remains unchanged, the higher the vertex add power, the greater the cone constant; when the central near vertex add power remains unchanged, the smaller the diameter, the smaller the cone constant.

[0073] Central distance vision zone diameter, near vision additional width and near vision focal power distribution

[0074] The central distance multifocal design category is used for eyes with larger pupils and for patients who require clear distance vision or high frequency distance vision tasks. The central distance zone must be smaller than the mesopic pupil size to benefit from a multifocal design. In some embodiments, the distance zone can be calculated based on a percentage of the mesopic pupil size using an equation that includes the mesopic pupil size, or can be based on an equation that is a function of the average of the scotopic and photopic pupil sizes and a factored equation that combines the mesopic pupil size with the range of scotopic to photopic pupil size measurements. In one embodiment, the rule that the algorithm uses to determine the central distance zone diameter can include a calculation equal to 70% of the mesopic pupil size. For example, if the mesopic pupil measures 4.4 mm, then the central distance zone diameter is equal to 3.08 mm, which is calculated based on 70% of a 4.4 mm mesopic pupil.

[0075] In some embodiments, the near addition portion of a central distance multifocal design is a ring. In some embodiments, an algorithm can calculate the width of the ring, which can be a function of the mesopic pupil size, the average of the scotopic and photopic pupil sizes, a factored equation combining the mesopic pupil size with the range of scotopic to photopic pupil size measurements, etc. The width of the ring can be determined by the mesopic pupil size, with the larger the pupil size, the wider the ring. In some embodiments, the minimum width of the ring can be 0.8 mm and the maximum width can be 2.0 mm. A linear transformation can be used, where for a 4.0 mm pupil, the width of the ring is 0.8 mm; for a pupil of 8.0 mm or larger, the width of the ring can be 2.0 mm. A certain spacing can be predetermined for the gain of the add power and the width of the add ring.

[0076] In some embodiments, the near add power of the annulus is equal to the near add power predicted by the algorithm in the previous step. In other embodiments, the vertex add power can be a multiple of the add power determined by the algorithm. In some embodiments, the vertex add power of the annulus can be 1.5 times the calculated add power. For example, the near add power calculated by the algorithm is +2.125, the near add annulus width is equal to 1.0 mm, and the add power of the near add annulus at the vertex is equal to +2.125*1.5, which is 3.188 D. The calculated conic constant must achieve a vertex power of 3.188 D at a distance of 0.5 mm from the local slope of the distance power at the central distance zone diameter of 3.08 mm calculated in the previous step.

[0077] After training, test data can be used to evaluate the performance of the machine learning algorithm. Test data can be collected in the same manner as training data. For example, a contact lens with a lens design, lens geometry, optical power, and optical power profile determined by the system can be placed in the patient's eye. The contact lens fit and visual function can be recorded and used to measure system performance. For example, if the contact lens fit and visual function are satisfactory, that data can be used to further refine the algorithm's parameters. Alternatively, if the contact lens fit and visual function are unsatisfactory, that resulting data can be used to further modify the algorithm.

[0078] Finally, the trained machine learning algorithm is applied to the newly acquired patient dataset, including clinical and lifestyle data, to determine lenses with lens designs and geometry, optical power, and optical power distribution, thereby providing a more customized multifocal contact lens for each eye of the new patient. By applying the trained machine learning algorithm to determine the individual lens characteristics, the system is able to minimize the impact of differences in practitioner qualifications and experience that would otherwise affect the lens selection process using traditional methods. In other words, patients receive the same multifocal contact lens parameter fitting results regardless of the eye care professional who examines and prescribes their lenses.

[0079] system

[0080] Figure 1 A system 100 for providing customized multifocal contact lenses to presbyopic patients according to embodiments disclosed herein is shown. The figure shows an example system 100 that can include a computing component 102 in communication with a network 140. The system 100 can also include one or more external resources 130 (e.g., an online repository of existing patient clinical records including patient information, including patient demographic information, patient physical characteristics, patient activities, and patient eye characteristics, clinical patient measurements, and multifocal zone parameters of multifocal contact lenses), a database 118, and a client computing device 120 in communication with the network 140. The external resources 130 can be located in a different physical or geographic location than the computing component 102.

[0081] like Figure 1 As shown, the computing component 102 or device can be, for example, a server computer, a controller, or any other similar computing component capable of processing data. Figure 1 In an example implementation of , computing component 102 includes a hardware processor 104 for executing one or more instructions stored in a machine-readable storage medium 105 including one or more computer program components.

[0082] The hardware processor 104 can be one or more central processing units (CPUs), semiconductor-based microprocessors, and / or other hardware devices suitable for retrieving and executing instructions stored in a computer-readable storage medium 105. The processor 104 can retrieve, decode, and execute instructions 106, 108, 110 to control a process or operation for determining a multifocal contact lens that provides an optimal lens-eye fit and optimal visual function for each eye of a patient with presbyopia. As an alternative or in addition to retrieving and executing instructions, the hardware processor 104 can include one or more electronic circuits including electronic components for performing the functions of one or more instructions, such as a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), or other electronic circuits.

[0083] A computer-readable storage medium, such as machine-readable storage medium 105, can be any electronic, magnetic, optical, or other physical storage device that contains or stores executable instructions. Thus, computer-readable storage medium 105 can be, for example, random access memory (RAM), non-volatile random access memory (NVRAM), electrically erasable programmable read-only memory (EEPROM), a storage device, and an optical disk. In some embodiments, machine-readable storage medium 105 can be a non-transitory storage medium, where the term "non-transitory" does not include transient propagating signals. As described in detail below, machine-readable storage medium 105 can be encoded with executable instructions, such as instructions 106, 108, and 110. As described above, hardware processor 104 can control processes / operations for facilitating design option customization by executing instructions 106, 108, and 110.

[0084] Hardware processor 104 can execute instructions 106 for training a machine learning algorithm using historical patient information and historical lens information. In some embodiments, the patient information can include patient demographic information, patient physical characteristics, patient activity information, and patient eye characteristics, clinical patient measurements, and multifocal zone parameters associated with a contact lens developed for the patient, so that the contact lens has an optimal lens-eye fit and optimal visual function for each eye of the presbyopic patient. In some embodiments, this information can be obtained from an online repository of existing patient clinical records and uploaded to database 118. In addition, hardware processor 104 can execute instructions 106 configured to also train the machine learning algorithm using a previous multifocal algorithm.

[0085] The hardware processor 104 may execute instructions 108 for applying the trained machine learning algorithm to new patient data. The hardware processor 104 may execute instructions 110 to generate a custom lens prescription based on the results of the component 106 .

[0086] When utilizing the system 100, a user (e.g., an eye care professional) can open and operate web applications, software applications, mobile applications, etc. on a client computing device 120, including but not limited to a computer, a client computing device 120 for displaying the lens selection interface described herein. For example, the eye care professional user can use an application 127 running on the client computing device 120 and communicating with the computing component 102 via the network 140.

[0087] Thus, the memory may store an application 127, which may include executable instructions that, when executed by the contact lens parameter prediction system 100, cause the contact lens parameter prediction system 100 to perform the actions described herein. The application 127 may be implemented as a module or component of another application. Furthermore, the application 127 may be implemented as an operating system extension, module, plug-in, etc. Furthermore, the application 127 may operate in a cloud-based computing environment. The application 127 may be executed within or as a virtual machine or virtual server managed in the cloud-based computing environment. Furthermore, the application 127, and even the multifocal contact lens parameter prediction system 100 itself, may reside in a virtual server running in the cloud-based computing environment, rather than being tied to one or more specific physical network computing devices. Furthermore, the application 127 may be executed in one or more virtual machines (VMs) executing on the multifocal contact lens parameter prediction system 100. Furthermore, in one or more embodiments of the present technology, the virtual machines running on the multifocal contact lens parameter prediction system 100 may be managed or supervised by a hypervisor.

[0088] Machine Learning Input Data and Training

[0089] In this particular example, the memory stores instructions for execution by the processor. Figure 2A Instructions for the indicated function. Figure 2A The elements of the process are provided in an arranged manner. However, it should be understood that one or more elements of the process can be performed in a different order, performed in parallel, or omitted entirely, etc. Other elements besides the elements provided can also be implemented, and in some examples, additional elements can be added to implement error correction functions if an anomaly occurs, etc.

[0090] exist Figure 2AIn block 210, the input data may include multiple data in different categories. The categories may include, for example, one or more patient-related information 212 and a prior multifocal algorithm 214. Specifically, patient-related information 212 may include patient demographic data 2121, patient lifestyle data 2122, and patient clinical information 2123. For example, patient clinical information 2123 may include subjective refraction, computerized autorefraction, ocular aberrations, accommodative amplitude, depth of field with reading add, corneal topography, ocular surface tomography, corneal curvature, eyelid position, palpebral fissure size, tear film break-up time, photopic pupil size, mesopic pupil size, scotopic pupil size, Kappa angle, deviation between the geometric center of the predicted contact lens and the geometric center of the pupil, aberrations while wearing the predicted contact lens, and sphero-cylindrical refraction while wearing the predicted contact lens. Similarly, patient demographic information 2121 may include any or all of the patient's height, age, gender, sitting height, and habitual head tilt. Similarly, patient lifestyle information 2122 may include any one or all of the patient's height, age, sex assigned at birth, sitting height, habitual head tilt angle, and responses to a patient questionnaire (e.g., a lifestyle questionnaire, a quality of life questionnaire), or responses to a structured patient interview, which may include information about the patient's eye and vision use in work and leisure activities, the number of hours the patient habitually spends in the corresponding activities, lighting conditions for the corresponding activities, viewing distance and working distance for the corresponding activities, and viewing height relative to the primary eye position in the posture position for the corresponding activities. Finally, patient-related information 212 may include multifocal contact lens design information 2124.

[0091] like Figure 2BAs shown, patient lens information 2124 may include lens design type 2141, lens optical geometry 2142, and central zone power 2143 and peripheral zone power 2144 of the lens. For example, lens design type 2141 may include a central distance design (i.e., a lens having a central zone for viewing distant objects) or a central near design (i.e., a lens having a central zone for viewing near objects). The lens design for the left and right eyes may be the same. In contrast, some patients may have one design in one eye (e.g., central distance for the right eye) and another design in the other eye (e.g., central near for the left eye). Lens geometry information 2142 may include, for example, the area of ​​the first zone of a multifocal contact lens, the distance and angular position from the first zone to the geometric center of the second zone, the overall diameter, the base curve radius, the posterior surface peripheral sagittal depth asymmetry, the anterior surface anti-rotation surface features, the thickness profile, or the surface modification. The central zone power 2143 and the peripheral zone power 2144 may include the power of each region (eg, the power of the central zone is different from the power of the peripheral zone). Finally, the distribution of the central zone power 2143 and the peripheral zone power 2144 may include the power distribution of each zone.

[0092] In addition, if Figure 2A As shown, the input data may include prior multifocal algorithms 214. These algorithms or rules may be developed based on empirical clinical data and may be used to determine parameters for a multifocal contact lens by observing the correspondence between relevant historical patient characteristics (e.g., age, height, measured reading distance) and visual function for a particular multifocal lens design (e.g., central distance vision or central near vision). This data, including prior multifocal algorithms 214, may be used to train a machine learning algorithm to predict which multifocal contact lens (i.e., a lens including a lens design and having lens geometry and optical power) will likely provide a personalized set of multifocal contact lens parameters for a new patient, given the new patient data (including clinical data and lifestyle data).

[0093] In some embodiments, success can be measured by clinical measurements and / or outcome assessments from questionnaires or structured interviews. The same clinical measurements and outcome assessments used to create the second lens for the patient's eye can be used to measure success. In some embodiments, success is measured by the ability to simultaneously report good distance, intermediate, and near vision, acceptable levels of flash, glare, ghosting, halos, or other visual noise, and acceptable visual function for the patient's individual lifestyle visual needs.

[0094] Clinical success can be measured by evaluating the lens-eye fit relationship, reported lens comfort, absence of inflammation or mechanical trauma signs, and measured visual function. Visual function assessment can be objective and subjective. Objective measurements of visual function include computerized automated refraction, aberrations of the lens-eye system, on-lens tomography, best contact lens-corrected visual acuity at each distance, etc. Subjective measurements can include subjective on-lens refraction at optical infinity, at an intermediate distance of approximately 0.5-2 m, and / or at a reading near distance of approximately 33-45 cm.

[0095] The feedback loop and machine learning may combine the raw input data, the first lens design output, the resulting evaluation of the first lens parameter set, the second lens parameter output, and the resulting evaluation of the second lens parameter set.

[0096] For example, as shown in Table 1, a center-distance or center-near designed lens type may include the following parameter determination rules that determine "addition power" and "distance power" based on relevant previous patient data.

[0097] Table 1

[0098]

[0099]

[0100] Table 2

[0101]

[0102] Table 3

[0103]

[0104] Table 4

[0105]

[0106]

[0107] Table 5

[0108]

[0109] Similarly, as shown in Tables 6 and 7, a center-distance designed lens type may include the following parameter determination rules that determine the "distance optical center position" and "addition chord position" based on relevant previous patient data.

[0110] Table 6

[0111]

[0112] Table 7

[0113]

[0114]

[0115] Finally, as shown in Tables 8 and 9, the central-near designed lens type may include the following parameter determination rules that determine the "addition diameter" and "addition position" based on relevant previous patient data.

[0116] Table 8

[0117]

[0118] Table 9

[0119]

[0120] Various machine learning models can be used. For example, machine learning models and techniques can include linear regression models, support vector machines (SVMs), classifiers, decision trees, neural networks, gradient boosting, and similar machine learning models and techniques. In some embodiments, a linear regression model can be used when there is a linear relationship between input features (e.g., measurements of eye curvature and corneal diameter) and the multifocal contact lens prescription. The model learns coefficients representing feature weights and predicts a prescription based on the input values. Similarly, decision tree-based algorithms such as random forests or gradient boosting trees can be used to handle nonlinear relationships between input features and prescriptions. These models can capture complex interactions and patterns in the data. In addition, support vector machines (SVMs) can be used to find the hyperplane that best separates the input feature space and predicts a multifocal contact lens prescription. They are effective for both linear and nonlinear relationships and can handle high-dimensional feature spaces. Finally, deep learning models such as multilayer perceptrons (MLPs) or convolutional neural networks (CNNs) can be applied to determine the multifocal contact lens prescription. These models can learn complex representations from the input features and are able to capture intricate relationships. Machine learning models can be pre-trained based on historical correspondences between input parameters and corresponding multifocal contact lens parameters. The input parameters may include the parameters described above, for example, the patient's age, sex assigned at birth, race, patient height, patient activity, and the size and shape of the patient's cornea. Once the machine learning model has been trained, new input parameters may be applied as input to the trained machine learning model. In response, the machine learning model may provide a multifocal contact lens prescription (i.e., lens parameter prediction) as output.

[0121] Some embodiments include training of a machine learning model. Training can be supervised, unsupervised, or a combination thereof and can be performed continuously between operations during the life of the system. Training can include generating a training set including the aforementioned input parameters and determining corresponding prescription predictions.

[0122] Training may include one or more second phases. The second phase may be after training and using the trained machine learning model and may include generating a second training set and using the second training set to train the trained machine learning model. During actual use of the machine learning model, the second training set may include inputs applied to the machine learning model and corresponding outputs generated by the machine learning model.

[0123] The second training phase may include specifying error estimates generated by the machine learning model and appending the specified error estimates to the second training set. Generating the second training set may also include appending input corresponding to the specified error estimates to the second training set.

[0124] Other data or components may be available without departing from the essence of this disclosure.

[0125] At block 220, the input data may be provided to the data interpolation module to initiate the data interpolation process. The data interpolation process may supplement the input data with additional information. In some examples, the data interpolation process may add information not provided in the initial clinical assessment. The data interpolation process may also add data later if the data was not initially retrievable.

[0126] At block 230 , training or inference functionality may be implemented.

[0127] At block 240, one or more clinical rules, demographic rules, or lifestyle rules may be added. For example, as described above and as Figure 2A The illustrated prior multifocal algorithm can be used to determine multifocal contact lens parameters based on relevant prior patient history characteristics (e.g., age, height, and measured reading distance) that correspond to observed visual function for a particular multifocal lens design (e.g., central distance vision or central near vision).

[0128] At block 250 , the outputs may be aggregated or combined.

[0129] At block 260 , an output or decision may be provided.

[0130] Applying machine learning methods

[0131] Figure 3 is a flow chart of an exemplary method for automatically predicting a multifocal contact lens parameter set for a presbyopic patient. Figure 3 The illustrative method provided in can be used by Figure 1 The multifocal contact lens parameter prediction system 100 is implemented.

[0132] Figure 3 The elements of the process are provided in an arranged manner. However, it should be understood that one or more elements of the process can be performed in a different order, performed in parallel, or omitted entirely, etc. Other elements besides the elements provided can also be implemented, and in some examples, additional elements can be added to implement error correction functions if an anomaly occurs, etc.

[0133] At block 310 , the method may receive a plurality of data including one or more of patient clinical information, patient demographic data, and patient lifestyle data.

[0134] At block 320 , the method may initiate a data interpolation process to supplement the plurality of data.

[0135] At block 330 , for each category of the plurality of data, the method may determine a classification value and a confidence level by applying a machine learning model to the category having the plurality of data as input.

[0136] Various machine learning models can be used. For example, machine learning models and techniques can include classifiers, decision trees, neural networks, gradient boosting, and similar machine learning models and techniques. Machine learning models can be pre-trained based on historical correspondences between input and output examples. Training can be supervised, unsupervised, or a combination thereof, and can be performed continuously between operations during the life of the system. The output of the model can be used to retrain the model, for example, to improve its accuracy. As discussed in Figure 2, the model can be trained and modified.

[0137] This article provides various illustrative examples of applying machine learning models to determine classification values ​​and confidence levels.

[0138] For example, the output of the machine learning model can determine, based on a confidence level, whether the first category of input data (e.g., pupil size and corrective prescription power) can determine multifocal zone parameters, such as the displacement or decentration, diameter, optical power, and optical power distribution of the multifocal zone. When the confidence level of this category exceeds a threshold for the first category, the multifocal zone parameters can be determined for the first category of data.

[0139] In another example, the output of the machine learning model can determine whether the second category of input data (e.g., patient physical characteristics) can be used to determine the multi-focal zone parameters of the second category based on the confidence level. When the confidence level of the second category exceeds a threshold of the second category, the multi-focal zone parameters can be determined for the second category of data.

[0140] In another example, the output of the machine learning model can determine whether a third category of input data (e.g., patient demographic information) can be used to determine multiple focal zone parameters for the third category based on a confidence level. When the confidence level for the category exceeds a threshold for the third category, multiple focal zone parameters can be determined for the third category of data.

[0141] In determining the weighted multi-focal zone parameters, data categories may be combined. For example, the three categories may have equal weights, such that when the classification value of the majority category is to determine all four multi-focal zone parameters and the corresponding confidence levels of these categories exceed the threshold value of each category, the weighted multi-focal zone parameter determination value that combines the classification values ​​for the plurality of data and the confidence levels of each category may also be equivalent to the multi-focal zone parameter determination value (e.g., corresponding to the majority). In another example of equal weights, when the classification value of the majority category is less than four multi-focal zone parameters (e.g., non-displacement or decentration of the multi-focal zone, diameter, optical power, and optical power distribution map) and the corresponding confidence levels of these categories exceed the threshold value of each category, the weighted multi-focal zone parameter determination value that combines the classification values ​​and confidence levels of each category for the plurality of data may also be equally reparable (e.g., corresponding to the majority).

[0142] Other weighted multi-focal zone parameter determination values ​​can be determined so that one determination value (e.g., classification value and confidence level) can have more weight or correspond to increased weight compared to other determination values. For example, when the confidence level exceeds a threshold, the first category (e.g., patient eye feature data) can be the default determination value for the multi-focal zone parameter. In other examples, the first category (e.g., patient eye feature data) can be the default determination value for the multi-focal zone parameter only when the confidence levels of the other categories do not exceed a threshold. In other examples, when the confidence level exceeds a threshold for the first category (e.g., patient eye feature data), an aggregation of a second category (e.g., patient physical characteristics) and a third category (e.g., patient demographic data) can be used. Various implementation details are possible.

[0143] The determined values ​​of the weighted multiple focus region parameters can be determined by voting. Voting can be used to adjust one or more thresholds corresponding to each data category, or to determine which input categories determine the multiple focus region parameters.

[0144] At block 340 , the method may determine a weighted multi-focus region parameter determination value that combines the classification value and the confidence level for each category of the plurality of data.

[0145] At block 350, the method may provide a graphical user interface (GUI) including display elements representing information associated with the weighted multiple focus region parameter determinations.

[0146] Computing Module

[0147] In one embodiment, where components, logic circuits, or engines of the technology are implemented in whole or in part using software, these software elements may be implemented to operate with computing circuitry or logic circuitry capable of performing the functions described therefor. Figure 4 An example computing module is shown in FIG. Various embodiments are described based on the example computing module 400. After reading this specification, it will be apparent to those skilled in the relevant art how to implement the technology using other logic circuits or architectures.

[0148] Figure 4 An example computing module 400 is shown, an example of which may be a processor / controller resident on a mobile device, or a processor / controller for operating a payment transaction device, which may be used to implement various features and / or functions of the systems and methods disclosed in this disclosure.

[0149] As used herein, the term module can describe a given functional unit that can be performed according to one or more embodiments of the present application. As used herein, a module can be implemented using any form of hardware, software, or a combination thereof. For example, one or more processors, controllers, application specific integrated circuits (ASICs), programmable logic arrays (PLAs), programmable array logic (PALs), complex programmable logic devices (CPLDs), field programmable gate arrays (FPGAs), logic components, software routines, or other mechanisms may be implemented to form a module. In implementation, the various modules described herein may be implemented as discrete modules, or the functions and features described may be partially or completely shared between one or more modules. In other words, after reading this specification, it will be apparent to those of ordinary skill in the art that the various features and functions described herein may be implemented in any given application and may be implemented in one or more separate or shared modules in various combinations and arrangements. Although the various features or elements of a function may be described separately or limited to separate modules, those of ordinary skill in the art will appreciate that these features and functions may be shared between one or more common software and hardware elements, and that such descriptions should not require or imply that separate hardware or software components are used to implement such features or functions.

[0150] In one embodiment, where components or modules of an application are implemented in whole or in part using software, these software elements may be implemented to operate together with a computing module or a processing module capable of performing the functions described therefor. Figure 4An example computing module is shown in FIG. Various embodiments are described based on the example computing module 400. After reading this specification, it will be apparent to those skilled in the relevant art how to use other computing modules or architectures to implement the application.

[0151] Now refer to Figure 4 , computing module 400 may represent computing or processing capabilities such as those found in desktop, laptop, notebook, and tablet computers; handheld computing devices (tablet computers, personal digital assistants (PDAs), smartphones, cell phones, and PDAs); mainframe computers, supercomputers, workstations, or servers; or any other type of specialized or general-purpose computing device that may be desirable or appropriate for a given application or environment. Computing module 400 may also represent computing capabilities embedded in or available to a given device. For example, computing modules may be found in other electronic devices such as digital cameras, navigation systems, cellular phones, portable computing devices, modems, routers, wireless application protocols (WAPs), terminals, and other electronic devices that may include some form of processing capabilities.

[0152] The computing module 400 may include, for example, one or more processors, controllers, control modules, or other processing devices, such as a processor 404. The processor 404 may be implemented using a general-purpose or special-purpose processing engine, such as a microprocessor, controller, or other control logic. In the illustrated example, the processor 404 is connected to a bus 402, although any communication medium may be used to facilitate interaction with other components of the computing module 400 or external communication. The bus 402 may also be connected to other components, such as a display 412, an input device 414, or a cursor controller 416, to help facilitate interaction and communication between the processor and / or other components of the computing module 400.

[0153] The computing module 400 may also include one or more memory modules, referred to herein as main memory 406. For example, a random access memory (RAM) or other dynamic storage may preferably be used to store information and instructions to be executed by the processor 404. The main memory 406 may also be used to store temporary variables or other intermediate information during the execution of instructions by the processor 404. The computing module 400 may also include a read-only memory ("ROM") 408 or other static storage device 410 coupled to the bus 402 for storing static information and instructions for the processor 404.

[0154] The computing module 400 may also include one or more information storage devices 410 in various forms, which may include, for example, a media drive and a storage unit interface. The media drive may include a drive or other mechanism that supports fixed or removable storage media. For example, a hard drive, a floppy disk drive, a tape drive, an optical drive, a CD or DVD drive (R or RW) or other removable or fixed media drive may be provided. Therefore, the storage medium may include, for example, a hard drive, a floppy disk, a magnetic tape, a cassette tape, an optical disk, a CD or DVD or other fixed or removable media that is read, written or accessed by a media drive. As shown in these examples, the storage medium may include a computer-usable storage medium in which computer software or data is stored.

[0155] In alternative embodiments, information storage device 410 may include other similar devices for enabling computer programs or other instructions or data to be loaded into computing module 400. These devices may include, for example, fixed or removable storage units and storage unit interfaces. Examples of these storage units and storage unit interfaces may include program cartridges and cartridge interfaces, removable memory (e.g., flash memory or other removable memory modules) and memory slots, memory card (PCMC IA) slots and memory cards, and other fixed or removable storage units and interfaces that enable software and data to be transferred from a storage unit to computing module 400.

[0156] The computing module 400 may also include a communication interface or network interface 418. The communication or network interface 418 may be used to enable software and data to be transferred between the computing module 400 and external devices. Examples of the communication or network interface 418 may include a modem or soft modem, a network interface (e.g., Ethernet, a network interface card, WiMedia, IEEE 802.XX, or other interface), a communication port (e.g., a USB port, an IR port, an RS232 port, a Bluetooth interface, or other port), or other communication interface. Software and data transferred via the communication or network interface 418 may typically be transmitted via signals, which may be electronic signals, electromagnetic signals (including optical signals), or other signals capable of being exchanged via a given communication interface. These signals may be provided to the communication interface 418 via a channel. The channel may transmit the signals and may be implemented using a wired or wireless communication medium. Some examples of a channel may include a telephone line, a cellular link, an RF link, an optical link, a network interface, a local or wide area network, and other wired or wireless communication channels.

[0157] In this application document, the terms "computer program medium" and "computer-usable medium" are used to refer generally to temporary or non-temporary media, such as memory 406, ROM 408, and storage unit interface 410. These and other various forms of computer program media or computer-usable media can involve transmitting one or more sequences of one or more instructions to a processing device for execution. These instructions contained on the medium are generally referred to as "computer program code" or "computer program product" (which can be grouped in the form of a computer program or other grouping). When executed, these instructions can cause computing module 400 to perform the features or functions of the present application discussed herein.

[0158] Various embodiments have been described with reference to specific exemplary features. However, it will be apparent that various modifications and changes may be made thereto without departing from the broader spirit and scope of the various embodiments set forth in the appended claims. Accordingly, the specification and drawings are to be regarded as illustrative rather than restrictive.

[0159] Although described above in terms of various exemplary embodiments and implementations, it should be understood that the various features, aspects, and functions described in one or more individual embodiments are not limited in their applicability to the specific embodiments described, but may be applied alone or in various combinations to one or more other embodiments of the present application, whether or not such embodiments are described as part of the described embodiments and whether or not such features are provided as part of the described embodiments. Accordingly, the breadth and scope of the present application should not be limited by any of the above-described exemplary embodiments.

[0160] Unless otherwise expressly stated, the terms and phrases used in this application and their variations should be understood as open and non-restrictive. As an example of the foregoing: the term "including" should be understood to mean "including but not limited to" and the like; the term "example" is used to provide an illustrative example of the item being discussed, rather than an exhaustive or limiting list thereof; the term "one" or "an" should be understood to mean "at least one", "one or more", and the like; and adjectives such as "conventional", "traditional", "normal", "standard", and "known", and terms of similar meaning should not be interpreted as limiting the items described to items available before a given time period or a given time, but should be understood to include conventional, traditional, normal, or standard technologies available or known now or at any time in the future. Similarly, when this application document refers to technologies that are obvious or known to a person of ordinary skill in the art, these technologies include technologies that are obvious or known to a person of ordinary skill in the art now or at any time in the future.

[0161] In some cases, the presence of broad words and phrases such as "one or more," "at least," "but not limited to," or other similar phrases should not be understood as intending or requiring a narrower context where such broad phrases may not be present. The use of the term "module" does not imply that the components or functionality described or defined as part of a module are all configured in a common package. In fact, any or all of the various components of a module, whether control logic or otherwise, may be combined in a single module package or maintained separately, and may also be distributed among multiple groups or module packages or across multiple locations.

[0162] In addition, the various embodiments described herein are described with reference to exemplary block diagrams, flow charts, and other illustrations. After reading this document, it will be apparent to those skilled in the art that the illustrated embodiments and their various alternatives may be implemented without limitation to the illustrated examples. For example, the block diagrams and their corresponding descriptions should not be construed as requiring a specific architecture or configuration.

Claims

1. A method for automatically determining multifocal lens parameters associated with a contact lens, the method comprising: receiving, by a lens parameter prediction system, a plurality of data including one or more patient eye characteristics, patient body characteristics, patient activity data, and patient demographic data; Initiating a data interpolation process to supplement the plurality of data; For each category of the plurality of data, determining a classification value and a confidence level of the category by applying a machine learning model with the plurality of data as input; determining a weighted multifocal lens parameter by combining the classification value and the confidence level for each category of the plurality of data; and A graphical user interface is provided that includes display elements, wherein the display elements represent information associated with the weighted multifocal lens parameters.

2. The method of claim 1, wherein the multifocal lens parameters include lens design and lens geometry.

3. The method of claim 1, wherein the multifocal lens parameters include a displacement or decentration of a zone of the contact lens, a diameter of the zone, a power of the zone, and a power profile of the zone.

4. The method of claim 2, wherein the lens design comprises a refractive design, a translational design, a diffractive design, a diffractive-refractive design, a pinhole aperture design, a spiral optical design, and / or an intelligent eye accommodation design.

5. The method of claim 2, wherein the lens geometry comprises an area of ​​a first zone of the contact lens, a distance and angular position from the first zone to the geometric center of a second zone, an overall diameter, a base curve radius, a posterior surface peripheral sagittal depth asymmetry, an anterior surface anti-rotation surface feature, a thickness profile, and / or surface modification measurements.

6. The method of claim 1 , wherein the patient eye characteristics include subjective refraction measurements, computerized autorefraction measurements, ocular aberration measurements, accommodative amplitude measurements, depth of field measurements under reading addition, corneal topography, ocular surface tomography, keratometric measurements, eyelid position measurements, palpebral fissure size, tear film break-up time, pupil size, and one or more of kappa angle or vertex normal measurements.

7. The method of claim 1, wherein the patient physical characteristics include the patient's height, arm length, age, sex assigned at birth, and race.

8. The method of claim 1, wherein the patient activity data comprises responses to interview questions used to determine the patient's eye usage.

9. The method of claim 8, wherein the patient's eye usage includes time spent at work and leisure activities, lighting conditions at work and leisure activities, viewing distance and working distance during work and leisure activities, and visual height of objects when performing work and leisure activities.

10. The method of claim 1, wherein the contact lens corrects presbyopia.

11. A contact lens for correcting refractive error in a patient's eye, comprising: front surface; posterior surface; the medium between the two; and a first zone having a first optical power and a second zone having a second optical power; wherein the first region has a first position and the second region has a second position; and wherein the first zone has a first optical power distribution, and the second zone has a second optical power distribution; and in, The first position and / or the second position of the first zone and / or the second zone, the first optical power and the second optical power, and at least one of the first optical power distribution and the second optical power distribution are automatically determined by a lens parameter prediction system based on multiple data, and the multiple data include one or more patient eye characteristics, patient body characteristics, patient activity data and patient demographic data.

12. A contact lens for correcting refractive errors in a patient's eye according to claim 11, wherein the medium of the contact lens is composed of a soft, hard or hybrid lens material.

13. A contact lens for correcting refractive errors in a patient's eye according to claim 11, wherein the medium of the contact lens partially or completely encapsulates a component.

14. A contact lens for correcting refractive error of a patient's eye according to claim 11, wherein the first optical power of the first zone corrects the refractive error of the eye.

15. A contact lens for correcting refractive error in a patient's eye according to claim 11, wherein the second optical power of the second zone corrects presbyopia.

16. A contact lens for correcting refractive error in a patient's eye according to claim 11, wherein the patient activity data comprises responses to interview questions used to determine the patient's eye usage.

17. A contact lens for correcting refractive errors in a patient's eye according to claim 16, wherein the patient's eye usage includes time spent in work and leisure activities, lighting conditions for work and leisure activities, viewing distance and working distance during work and leisure activities, and visual height of objects viewed while performing work and leisure activities.

18. The contact lens for correcting refractive error in a patient's eye according to claim 11, wherein the patient's eye characteristics include one or more of subjective refraction measurement, computerized autorefraction measurement, ocular aberration measurement, accommodative amplitude measurement, depth of field measurement under reading addition, corneal topography, ocular surface tomography, keratometric measurement, eyelid position measurement, palpebral fissure size, tear film break-up time, pupil size, and kappa angle or vertex normal measurement.

19. A contact lens for correcting refractive error in a patient's eye according to claim 11, wherein the patient's physical characteristics include the patient's height, arm length, age, sex assigned at birth, and race.

20. The contact lens for correcting refractive errors of a patient's eye according to claim 11, wherein the lens design is automatically determined by the lens parameter prediction system, and the lens design includes a refractive design, a translational design, a diffractive design, a diffractive-refractive design, a pinhole aperture design, a spiral optical design and / or an intelligent eye adjustment design.

Images