Intraocular lens selection based on predicted subjective outcome scores

Abstract

A system and method for selecting an intraocular lens for implantation in an eye includes a controller having a processor and a tangible, non-transitory memory having instructions recorded thereon. The controller is configured to selectively execute a machine learning model trained using a training dataset. Execution of the instructions by the processor causes the controller to acquire pre-operative objective data for the patient, including one or more anatomical ocular measurements. The controller is configured to acquire pre-operative questionnaire data for the patient, including at least one personality trait. The pre-operative objective data and the pre-operative questionnaire data are input as respective inputs to the machine learning model. A predicted subjective outcome score for the patient is generated as an output of the machine learning model. The intraocular lens is selected based in part on the predicted subjective outcome score.

Description

[Technical Field]

[0001] FIELD OF THE DISCLOSURE The present disclosure relates generally to systems and methods for selecting an intraocular lens for implantation in an eye. [Background technology]

[0002] Generally, the human crystalline lens is transparent, allowing light to pass through easily. However, various factors can cause areas within the lens to become cloudy and opaque, adversely affecting the quality of vision. This condition can be treated with cataract surgery, in which an artificial lens is selected for implantation within the patient's eye. In fact, cataract surgery is commonly performed worldwide. Today, various types of intraocular lenses are available, and it is not always clear what may be the best choice for a particular patient. Summary of the Invention [Means for solving the problem]

[0003] Disclosed herein are systems and methods for selecting an intraocular lens for implantation in a patient's eye. The system includes a controller having a processor and a tangible, non-transitory memory having instructions recorded thereon. The controller is configured to selectively execute at least one machine learning model (hereinafter "at least one" is omitted). The machine learning model is trained using a training dataset.

[0004] Execution of the instructions by the processor causes the controller to acquire pre-operative objective data for the patient, including one or more anatomical ocular measurements. The controller is configured to acquire pre-operative questionnaire data for the patient, including at least one personality trait. The pre-operative objective data and the pre-operative questionnaire data are input as respective inputs to a machine learning model. A predicted subjective outcome score for the patient is generated as an output of the machine learning model. An intraocular lens is selected based in part on the predicted subjective outcome score.

[0005] The machine learning model may include a neural network. The patient's personality traits may be represented as at least one of a numerical scale of agreeableness or a binary outcome, where the binary outcome is either predominantly agreeable or predominantly disagreeable. The preoperative questionnaire data may further include a patient lifestyle needs assessment.

[0006] The integrated diagnostic device may be configured to acquire pre-operative objective data. The pre-operative objective data may further include refractive eye measurements and physiological eye measurements. The training data set includes respective history sets consisting of respective pre-operative objective data, respective pre-operative personality data, respective intra-operative data, respective post-operative objective data, and respective subjective outcome data. The system may include a data management module configured to collect the respective history sets from a plurality of electronic medical record units and deliver the respective history sets to at least one machine learning module. Each subjective outcome data in each history set may include a numerical satisfaction scale.

[0007] The controller may be configured to quantify a correlation of each post-operative objective data in each historical set to each subjective outcome score and identify each post-operative objective data that is most strongly correlated with each subjective outcome score. The controller may be configured to identify and screen each historical set having at least one variable in each pre-operative objective data that is consistent with a predetermined confounding parameter.

[0008] The above and other features and advantages of the present disclosure will become readily apparent from the following detailed description of the best mode for carrying out the disclosure, when read in conjunction with the accompanying drawings. [Brief explanation of the drawings]

[0009] [Figure 1]FIG. 1 is a schematic diagram of a system for selecting an intraocular lens for implantation in an eye, the system having a controller. [Figure 2] FIG. 2 is a schematic flow diagram of a method that can be performed by the controller of FIG. [Figure 3] FIG. 3 is a schematic example of a neural network that can be implemented by the controller of FIG. DETAILED DESCRIPTION OF THE INVENTION

[0010] Referring to the drawings, wherein like reference numerals refer to like components, FIG. 1 schematically illustrates a system 10 for selecting an intraocular lens L for implantation in a patient P. As described below, the system 10 utilizes both objective and subjective data to optimize the selection process. Illustrated in FIG. 1 are examples of a first intraocular lens 12 and a second intraocular lens 22. The first intraocular lens 12 and the second intraocular lens 22 are multifocal lenses in the illustrated example, although it will be understood that any type of intraocular lens L available to one skilled in the art may be employed.

[0011] Referring to FIG. 1 , a first intraocular lens 12 includes an optic zone 14 contiguous with one or more support structures 16. The optic zone 14 may include an apodized diffractive multifocal zone 18 and an outer distance zone 20, and the first intraocular lens 12 is configured to provide good vision over a wide distance range. Referring to FIG. 1 , a second intraocular lens 22 includes an optic zone 24 contiguous with one or more support structures 26. The optic zone 24 may include an apodized diffractive multifocal zone 28, an outer distance zone 30, and a central distance zone 32. The second intraocular lens 22 may be configured to provide sharper distance vision and improved intermediate vision compared to the first intraocular lens 12.

[0012] Alternatively, the intraocular lens L may be a monofocal lens. The intraocular lens L may also be a containing lens having an internal cavity filled with a fluid, which is movable to change the thickness (and power) of the intraocular lens L. It will be appreciated that the intraocular lens L may take many different forms and may include multiple and / or alternative components.

[0013] 1, system 10 includes a controller C having at least one processor 36 and at least one memory 38 (or non-transitory tangible computer-readable storage medium) having instructions recorded thereon for performing a method 100 for selecting an intraocular lens L for a patient P. Method 100 is shown and described below with reference to FIG.

[0014] 1 , the controller C is specifically programmed to selectively execute one or more machine learning models 40, such as a first machine learning model 42 and a second machine learning model 44. The machine learning models 40 may be embedded in the controller C. The machine learning models 40 may also be stored elsewhere and accessible to the controller C. The machine learning models 40 may be configured to find parameters, weights, or structures that minimize a respective cost function.

[0015] Referring to FIG. 1 , the machine learning model 40 is trained using one or more training datasets from multiple facilities 50, such as a first facility 52, a second facility 54, and a third facility 56, which may be clinical facilities located around the world. A controller C can communicate with the multiple facilities 50 via a first network 58. The training dataset includes respective history sets for multiple patients. As described below, each history set includes each patient's respective preoperative objective data, respective preoperative personality data, respective intraoperative data, respective postoperative objective data, and respective subjective outcome data. The training dataset may be stratified based on demographic data, patients with similar eye size, or other health status factors. Each of the multiple facilities 50 may include an integrated diagnostic device 60 configured to acquire the preoperative objective data.

[0016] 1 , the system 10 may include a data management module 62 having a computerized data management system capable of storing information from electronic medical records of each of the multiple facilities 50. The data management module 62 is configured to collect respective history sets from the multiple facilities 50 and provide them to the controller C. The data management module 62 may include a cloud unit 64 and / or a remote server 66 configured to share data across all clinical facilities employing the system 10. The cloud unit 64 may comprise one or more servers hosted on the Internet for data storage, management, and processing. The remote server 66 may be a private or public information source maintained by an organization such as, for example, a research institute, a company, a university, and / or a hospital.

[0017] Referring to FIG. 1 , a patient P may be associated with a clinic 70. A controller C may be configured to receive and transmit communications with the clinic 70 via a user interface 72. The user interface 72 may be installed on a smartphone, laptop, tablet, desktop, or other electronic device that may be operated by a care provider at the clinic 70 using, for example, a touchscreen interface or an I / O device such as a keyboard or mouse. The user interface 72 may be a mobile application. Mobile application (“app”) circuits and components available to those skilled in the art may be employed. The user interface 72 may include an integrated processor 74 and an integrated memory 76. The user interface 72 may communicate with the controller C via a second network 78 to access data within the controller C.

[0018] 1 , the user interface 72 may include multiple modules, such as a first module 80, a second module 82, and a third module 84. In one example, the first module 80 and the second module 82 are configured to provide input factors (pre-operative objective data and pre-operative questionnaire data, respectively) to a common or different machine learning model 40. In another example, the third module 84 is configured to obtain the output (predicted subjective outcome score) of the machine learning model 40. The user interface 72 may include a database 86 for storing and comparing the output (predicted subjective outcome score) of different types of intraocular lenses L.

[0019] 1 , the first network 58 and / or the second network 78 may be a wireless local area network (LAN) that connects multiple devices in a wirelessly distributed manner, a wireless metropolitan area network (MAN) that connects several wireless LANs, or a wireless wide area network (WAN) that covers a large geographic area such as a nearby city or town. Other types of connections may also be employed. The first network 58 and / or the second network 78 may be a bus implemented in various ways, such as a serial communication bus in the form of a local area network. The local area network may include, but is not limited to, a Controller Area Network (CAN), a Controller Area Network with Flexible Data Rate (CAN-FD), Ethernet, Bluetooth, Wi-Fi, and other data connection topologies.

[0020] Referring now to Figure 2, there is shown a flow diagram of a method 100 executable by the controller C of Figure 1. The method 100 does not have to be applied in the particular order described herein, and some blocks may be omitted. The memory M may store a set of controller-executable instructions, and the processor P may execute the set of controller-executable instructions stored in the memory M.

[0021] 2, the controller C is configured to collect one or more training data sets from the remote server 40, for example, via the long-distance network 44. The training data sets include respective history sets consisting of respective pre-operative objective data (block 102A), respective pre-operative personality data (block 102B), respective intra-operative data (block 102C), respective post-operative objective data (block 102D), and respective subjective outcome data (block 102E). In block 104, the controller C is configured to calibrate or train the machine learning model 40 using the training data sets from block 102.

[0022] For block 102A, each pre-operative objective data may include anatomical ocular measurements (e.g., ocular length, corneal shape and thickness, lens position and thickness, etc.), refractive ocular measurements (e.g., classical refraction values, wavefront aberration measurements), and physiological ocular measurements (e.g., intraocular pressure, tear film health values, etc.). Each pre-operative objective data (block 102A) may include additional visual function measurements (e.g., photopic / mesopic visual acuity, contrast sensitivity, near visual acuity, etc.). In addition, controller C may be configured to identify and screen each history set having at least one variable in each post-operative objective data that is consistent with a predetermined confounding parameter. For example, if the confounding parameter is previous eye surgery, the corresponding data set may be excluded from the training data set.

[0023] For block 102B, each preoperative personality data may include a vision needs assessment or lifestyle requirements (predominant activity, e.g., needlepoint vs. fishing) and personality traits. Standardized assessments may be used to obtain consistent data across patients and sites. In one example, the Big Five Factor model of personality types, sometimes known as McCrae and Costa, may be employed. The Big Five Factor model posits that the traits of openness, conscientiousness, extraversion, agreeableness, and neuroticism (or emotional stability) form the basis of people's personalities (see McCrae, R., Costa, P., Personality in Adulthood: A Five-Factor Theory Perspective, Guilford Press, New York City (2003)). In one example, personality traits may be expressed as how agreeable a person is on a numerical scale of agreeableness, e.g., a scale of 1 to 10. In another example, personality traits may be expressed as binary outcomes, i.e., either agreeableness predominates or disagreeableness predominates.

[0024] For block 102C, each intraoperative data (block 102C) may include information related to the actual treatment performed. This information may be electronically captured and provided within the data management module 62. Examples of intraoperative data include, but are not limited to, the type of refractive surgery procedure performed, the model of the implanted intraocular lens, and its prescription. The intraoperative data may include intraoperative aberrometry measurements. The intraoperative data may further include surgical machine settings and parameters, such as procedure time, operating room temperature, total phaco power consumed to emulsify the original lens, the duration for which phaco energy was applied, and effective phaco time (as the product of phaco time multiplied by average phaco power). The intraoperative data may further include the type of delivery device used to implant the intraocular lens, the presence or absence of occlusion disruption, the amount and extent of occlusion disruption, and whether ancillary devices (such as capsule hooks) were employed. The intraoperative data may also include an intraoperative grade of the original lens's nuclear hardness, which may be graded according to the lens's opacity classification.

[0025] For block 102D, each postoperative objective data may include objective, measurable information obtained postoperatively for each patient in the training dataset. Each postoperative objective data may include anatomical ocular measurements (e.g., ocular length, corneal shape and thickness, lens position and thickness, etc.), refractive ocular measurements (e.g., classical refraction, wavefront aberration measurements), physiological ocular measurements (e.g., intraocular pressure, tear film health, etc.), and visual function measurements (e.g., photopic / mesopic visual acuity, contrast sensitivity, near visual acuity, etc.).

[0026] For block 102E, each subjective outcome data in each historical set may include one or more numerical satisfaction scales reflecting satisfaction with postoperative visual results. Patient satisfaction with surgical results may be captured at one or more specific time periods (e.g., one month postoperatively and three months postoperatively). In one example, a single overall satisfaction scale is employed based on the following question: "On a scale of 1 to 5 (5 being best), how satisfied are you with your current vision?" In another example, separate satisfaction scales may be employed for near vision, distance vision, night / twilight vision, "outdoor sports vision" (e.g., when playing golf), and overall satisfaction.

[0027] Controller C may be configured to quantify the correlation of each postoperative objective data in each history set to each subjective outcome score and identify each postoperative objective data that most strongly correlates with each subjective outcome score. In other words, system 10 looks at the objective postoperative measurements and evaluates how much they influence the questionnaire responses (each subjective outcome data). This provides two technical advantages. First, it allows for the identification of objective postoperative measurements that drive patient satisfaction / dissatisfaction, and second, it allows for the screening of patients with confounding outcome parameters. For example, controller C may be configured to screen each history set if each postoperative objective data exceeds a threshold, e.g., if the postoperative refraction exceeds half a diopter from the intended one.

[0028] 2, block 106, the controller C is configured to acquire pre-operative objective data of the patient P at the clinic 70. The pre-operative objective data may include anatomical ocular measurements (e.g., ocular length, corneal shape and thickness, lens position and thickness, etc.), refractive ocular measurements (e.g., classical refraction values, wavefront aberration measurements), physiological ocular measurements (e.g., intraocular pressure, tear film health values, etc.), and visual function measurements (e.g., photopic / mesopic visual acuity, contrast sensitivity, near visual acuity, etc.).

[0029] According to block 108 of FIG. 2 , the controller C is configured to obtain preoperative questionnaire data for the patient P, including at least one personality trait. As described above, a standardized method may be used to assess the personality traits of the patient P. In one example, the personality traits of the patient P may be expressed as a numerical scale of agreeableness. In another example, the personality traits may be expressed as a binary outcome, such that either agreeableness or disagreeableness predominates for the patient P. The preoperative questionnaire data may further include a lifestyle needs assessment for the patient P.

[0030] According to block 110 of FIG. 2, the method 100 includes inputting the pre-operative objective data and the pre-operative questionnaire data as respective inputs to the machine learning model 40 and executing the machine learning model 40. According to block 112 of FIG. 2, the controller C is configured to generate, as an output of the machine learning model 40, a predicted subjective outcome score for the patient P for the first intraocular lens 12. Blocks 102-112 may be repeated to obtain predicted subjective outcome scores for the patient P for the second intraocular lens 22 and other types of intraocular lenses. Alternatively, the training dataset may include multiple types of intraocular lenses, and the type of intraocular lens is incorporated as an element of block 102C.

[0031] According to block 114, the method 100 includes selecting an appropriate intraocular lens L based in part on a comparison of the predicted subjective outcome scores for the first intraocular lens 12, the second intraocular lens 22, and other lenses. For example, if the predicted subjective outcome scores are 85% for the first intraocular lens 12 and 30% for the second intraocular lens 22, bilateral implantation of the first intraocular lens 12 may be optimal. If the predicted subjective outcome scores are 55% for the first intraocular lens 12 and 60% for the second intraocular lens 22, refractive results may be optimized by a "blended" solution, i.e., implanting the second intraocular lens 22 in the dominant eye and the first intraocular lens 12 in the non-dominant eye.

[0032] System 10 may be configured to be "adaptive" and periodically updated after collection of additional data for the training dataset. In other words, machine learning model 40 may be configured to be an "adaptive machine learning" algorithm that is not static but improves after additional training datasets are collected. Machine learning model 40 of FIG. 1 may be configured to find parameters, weights, or structures that minimize respective cost functions and may incorporate respective regression models. Machine learning model 40 of FIG. 1 may include a neural network, an example of which is shown in FIG. 3.

[0033] Referring to FIG. 3, neural network 200 is a feedforward artificial neural network with at least three layers, including an input layer 201, at least one hidden layer 220, and an output layer 240. Each layer is composed of a respective node N configured to perform an affine transformation of a linear sum of inputs. Each node N is characterized by a respective bias and a respective weighted link. The parameters of each node N may be independent of others, i.e., characterized by a unique set of weights. Input layer 201 may include a first input node 202, a second input node 204, a third input node 206, a fourth input node 208, a fifth input node 210, and a sixth input node 212. Each node N in input layer 201 receives inputs, normalizes them, and forwards them to a respective node N in hidden layer 220.

[0034] Referring to FIG. 3 , hidden layer 220 may include first hidden node 222, second hidden node 224, third hidden node 226, fourth hidden node 228, and fifth hidden node 230. Each node N in a subsequent layer computes a linear combination of the outputs of the previous layer. A network with three layers forms an activation function f(x) = f(3)(f(2)(f(1)(x))). The activation function f may be linear for each node N in output layer 240. The activation function f may be a sigmoid in hidden layer 220. A linear combination of sigmoids may be used to approximate a continuous function that characterizes the output vector y. Patterns recognized by neural network 200 may be translated or converted into numerical form and organized into vectors or matrices.

[0035] The machine learning model 40 can match an input vector x to an output vector y by using a deep learning map to learn an activation function f such that f(x) maps to y. The training process allows the machine learning model 40 to correlate the appropriate activation function f(x) to transform the input vector x into the output vector y. For example, in a simple linear regression model, two parameters are learned: bias and slope. The bias is the level of the output vector y when the input vector x is set to 0, and the slope is the predicted rate of increase or decrease of the output vector y for each unit increase in the input vector x. Once the machine learning model 40 is trained, an estimate of the output vector y can be calculated with new values ​​of the input vector x.

[0036] In summary, the system 10 and method 100 utilize both objective and subjective parameters to optimize the selection process for the intraocular lens L. The system 10 and method 100 provide objective guidance to both the clinician and the patient based on detailed preoperative patient information and a database of previous cases that incorporates both subjective and objective information as well as details of the surgical procedure applied in each case.

[0037] The controller C of FIG. 1 includes computer-readable media (also referred to as processor-readable media), including non-transitory (e.g., tangible) media that participate in providing data (e.g., instructions) that can be read by a computer (e.g., by a computer processor). Such media may take many forms, including, but not limited to, non-volatile and volatile media. Non-volatile media may include, for example, optical or magnetic disks and other persistent memory. Volatile media may include, for example, dynamic random access memory (DRAM), which may constitute primary storage. Such instructions may be transmitted over one or more transmission media, including coaxial cables, copper wire, and optical fiber, including the wires that comprise a system bus coupled to the computer's processor. Some forms of computer-readable media include, for example, floppy disks, flexible disks, hard disks, magnetic tape, or other magnetic media; CD-ROMs, DVDs, or other optical media; punch cards, paper tape, or other physical media with patterns of holes; RAM, PROMs, EPROMs, Flash EEPROMs, or other memory chips or cartridges; or other computer-readable media.

[0038] The lookup tables, databases, data repositories, or other data stores described herein may include various types of mechanisms for storing, accessing, and retrieving various types of data, including a hierarchical database, a set of files in a file system, a proprietary application database, a relational database management system (RDBMS), etc. Each such data store may be contained within a computing device employing a computer operating system such as those described above, or may be accessed over a network in one or more of a variety of ways. The file system may be accessible from the computer operating system and may include files stored in various formats. The RDBMS may employ a Structured Query Language (SQL) in addition to a language for creating, saving, editing, and executing stored procedures, such as the PL / SQL language described above.

[0039] While the detailed description and drawings or figures support and explain the present disclosure, the scope of the present disclosure is defined solely by the claims. While the best mode and some alternative embodiments for carrying out the claimed disclosure have been described in detail, various alternative designs and embodiments exist for carrying out the disclosure defined in the appended claims. Furthermore, the features of the embodiments shown in the drawings or described herein should not necessarily be understood as independent embodiments. Rather, each of the characteristics described in one of the example embodiments can be combined with one or more other desirable characteristics from other embodiments, resulting in other embodiments not described in words or with reference to the drawings. Accordingly, such other embodiments are encompassed within the scope of the appended claims.

Claims

1. 1. A system for selecting an intraocular lens for implantation in a patient's eye, comprising: a controller having a processor and a tangible non-transitory memory having instructions stored therein; the controller is configured to selectively execute at least one machine learning model, the at least one machine learning model being trained using a training dataset; Execution of the instructions by the processor causes the controller to: obtaining preoperative objective data for the patient, including one or more anatomical ocular measurements; obtaining preoperative questionnaire data from the patient, the questionnaire data including at least one personality trait; inputting the patient's pre-operative objective data and the patient's pre-operative questionnaire data as respective inputs to the at least one machine learning model and generating a predicted subjective outcome score for the patient as an output of the at least one machine learning model; The system causes the intraocular lens to be selected based in part on a predicted subjective outcome score for the patient.

2. The system of claim 1 , wherein the at least one machine learning model comprises a neural network.

3. 10. The system of claim 1, wherein the at least one personality trait of the patient is represented as at least one of a numerical scale of agreeableness or a binary outcome, the binary outcome being either predominantly agreeable or predominantly disagreeable.

4. The system of claim 1 , wherein the preoperative questionnaire data further comprises a lifestyle needs assessment of the patient.

5. an integrated diagnostic device configured to acquire the preoperative objective data; The system of claim 1 , wherein the pre-operative objective data further comprises refractive eye measurements and physiological eye measurements.

6. 2. The system of claim 1, wherein the training data set includes respective history sets consisting of respective pre-operative objective data, respective pre-operative personality data, respective intra-operative data, respective post-operative objective data, and respective subjective outcome data.

7. The system of claim 6 , further comprising a data management module accessible to the controller, the data management module configured to collect the respective history sets from a plurality of electronic medical record units.

8. The system of claim 6 , wherein the respective subjective outcome data in the respective historical set comprises a numerical satisfaction scale.

9. The controller: quantifying the correlation of each of the post-operative objective data to each of the subjective outcome data in each of the historical sets; and The system of claim 6 , configured to identify the respective post-operative objective data that most strongly correlates with the respective subjective outcome data.

10. The controller:

7. The system of claim 6, configured to identify and screen the respective history sets having at least one variable in the respective preoperative objective data that is consistent with a predetermined confounding parameter.

11. 1. A method for selecting an intraocular lens for implantation in an eye, comprising: receiving, via the controller, pre-operative objective data of the patient, including one or more anatomical ocular measurements; receiving, via the controller, preoperative questionnaire data of the patient, the questionnaire data including at least one personality trait; running, via the controller, at least one machine learning model using the patient's preoperative objective data and the patient's preoperative questionnaire data as respective inputs, wherein the at least one machine learning model has been trained using a training dataset; and generating a predicted subjective outcome score for the patient as an output of the at least one machine learning model; selecting the intraocular lens based in part on the patient's predicted subjective outcome score.

12. 12. The method of claim 11, wherein the at least one personality trait of the patient is expressed as at least one of a numerical scale of agreeableness or a binary outcome, the binary outcome being either predominantly agreeableness or predominantly disagreeableness.

13. The method of claim 11 , wherein the preoperative questionnaire data further comprises a lifestyle needs assessment of the patient.

14. an integrated diagnostic device configured to acquire the preoperative objective data; The method of claim 11 , wherein the pre-operative objective data further comprises refractive eye measurements and physiological eye measurements.

15. 12. The method of claim 11, wherein the training data set includes respective history sets consisting of respective pre-operative objective data, respective pre-operative personality data, respective intra-operative data, respective post-operative objective data, and respective subjective outcome data.

16. collecting the respective history sets from a plurality of electronic medical record units; and The method of claim 15 , further comprising a data management module configured to deliver the respective history sets to the at least one machine learning model.

17. The method described in claim 15, wherein each of the subjective outcome data in each of the history sets includes a numerical satisfaction scale. evaluating, via the controller, a correlation of each of the postoperative objective data in each of the history sets to each of the subjective outcome data; 16. The method of claim 15, further comprising: identifying, via the controller, the respective post-operative objective data that most strongly correlates with the respective subjective outcome data.

Images