Predicting optimal treatment regimens for patients with neovascular age-related macular degeneration (NAMD) using machine learning

Abstract

A method and system for predicting a selected treatment regimen for a subject. Baseline data for a subject diagnosed with neovascular age-related macular degeneration (nAMD) is received. A plurality of predictor inputs for an outcome predictor are formed using the baseline data and the regimen data for the plurality of treatment regimens. The plurality of predictor inputs includes a different predictor input for each of the plurality of treatment regimens. A plurality of treatment scores are generated for the plurality of treatment regimens via a set of outcome predictors using the plurality of predictor inputs. One of the plurality of treatment regimens is selected as the selected treatment regimen for the subject based on the plurality of treatment scores.

Description

[Technical field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This application claims priority to U.S. Provisional Patent Application No. 63 / 415,949, entitled "Predicting Optimal Treatment Regimen for Neovascular Age-Related Macular Degeneration (NAMD) using Machine Learning," filed on October 13, 2022, U.S. Provisional Patent Application No. 63 / 330,753, entitled "Predicting Optimal Treatment Regimen for Neovascular Age-Related Macular Degeneration (NAMD) using Machine Learning," filed on April 13, 2022, and U.S. Provisional Patent Application No. 63 / 291,275, entitled "Predicting Optimal Treatment Regimen for Neovascular Age-Related Macular Degeneration (NAMD) using Machine Learning," filed on December 17, 2021, each of which is incorporated by reference in its entirety into this specification.

[0002] Field This description is generally directed to predicting the optimal therapeutic regimen for use in a subject diagnosed with neovascular age-related macular degeneration (nAMD). More specifically, this description provides a method and system for selecting one of a variety of potential therapeutic regimens for use in a subject with nAMD based on a treatment score of the therapeutic regimen generated using a machine learning model. [Background technology]

[0003] Age-related macular degeneration (AMD) is a disease that affects the central area of ​​the retina of the eye, called the macula. AMD is the leading cause of vision loss in subjects over the age of 50. Neovascular AMD (nAMD) is one of two advanced stages of AMD. In nAMD, new, abnormal blood vessels grow uncontrollably under the macula and can leak fluid from the blood into the retina. This type of growth can cause swelling, bleeding, fibrosis, other problems, or a combination thereof. Furthermore, the fluid that leaks into the retina can immediately distort a subject's vision and over time can damage the retina itself, for example, by causing loss of retinal photoreceptors. The fluid can also cause the macula to separate from its base, resulting in severe and rapid vision loss.

[0004] Treatment of nAMD typically involves anti-vascular endothelial growth factor (anti-VEGF) therapy (e.g., anti-VEGF drugs such as ranibizumab). The retinal response to such treatment is at least partially subject-specific, such that different subjects may respond differently to the same type of anti-VEGF drug. Faricimab is a new treatment for nAMD and is a monoclonal antibody therapy. Both anti-VEGF therapy and treatment with monoclonal antibodies such as faricimab are typically administered by intravitreal injection, which can be expensive in some cases and associated with side effects or complications (e.g., redness, eye pain, infection, blindness, etc.). The number or frequency of injections can also be burdensome for patients and lead to reduced disease control. Summary of the Invention

[0005] In one or more embodiments, a method is provided for predicting a selected treatment regimen for a subject with neovascular age-related macular degeneration (nAMD). Baseline data of a subject diagnosed with nAMD is received. A plurality of predictor inputs for the outcome predictor are formed using baseline data and regimen data for the plurality of treatment regimens. The plurality of predictor inputs include different predictor inputs for each of the plurality of treatment regimens, and the outcome predictor includes at least one machine learning model. A plurality of treatment scores for the plurality of treatment regimens are generated via the outcome predictor using the plurality of prediction inputs. Based on the plurality of treatment scores, one of the plurality of treatment regimens is selected as the selected treatment regimen for the subject.

[0006] In one or more embodiments, a method is provided for predicting a selected treatment regimen for a subject with nAMD. Clinical data and imaging data can be received for a subject diagnosed with nAMD for a baseline time point. The imaging data is used to generate retinal feature data. The retinal feature data includes at least one of pathology-related features, layer-related volumetric features, or layer-related thickness. A plurality of treatment scores for the plurality of treatment regimens are generated via an outcome predictor that includes at least one machine learning model. Based on the plurality of treatment scores, one of the plurality of treatment regimens is selected as the selected treatment regimen for the subject.

[0007] In one or more embodiments, a system for predicting a selected treatment regimen for a subject diagnosed with neovascular age-related macular degeneration (nAMD) is provided. The system includes a memory including a machine-readable medium including machine-executable code, and a processor coupled to the memory. The processor is configured to execute the machine-executable code to cause the processor to receive baseline data for the subject diagnosed with nAMD. The processor is configured to execute the machine-executable code to cause the processor to form a plurality of predictor inputs for an outcome predictor using the baseline data and regimen data for the plurality of treatment regimens. The plurality of predictor inputs includes a different predictor input for each of the plurality of treatment regimens. The outcome predictor includes at least one machine learning model. The processor is configured to execute the machine-executable code to cause the processor to generate a plurality of treatment scores for the plurality of treatment regimens using the plurality of predictor inputs via the outcome predictor. The processor is configured to execute the machine-executable code to cause the processor to select one of the plurality of treatment regimens as a selected treatment regimen for the subject based on the plurality of treatment scores.

[0008] In some embodiments, a system is provided that includes one or more data processors and a non-transitory machine-readable storage medium that includes instructions that, when executed on the one or more data processors, cause the one or more data processors to perform some or all of the methods disclosed herein.

[0009] In some embodiments, a computer program product is provided that is tangibly embodied in a non-transitory machine-readable storage medium and includes instructions configured to cause one or more data processors to perform some or all of the methods disclosed herein. [Brief description of the drawings]

[0010] The present disclosure is described in conjunction with the accompanying drawings, in which:

[0011] [Figure 1] FIG. 1 is a block diagram of a prediction system 100 according to various embodiments.

[0012] [Diagram 2] 2 is a block diagram of the outcome prediction system 110 from FIG. 1 according to one or more embodiments.

[0013] [Diagram 3] FIG. 1 is a schematic diagram of an exemplary workflow for selecting a treatment regimen for use in a subject with neovascular age-related macular degeneration (nAMD), according to one or more embodiments.

[0014] [Figure 4] 1 is a flowchart of a process for selecting a treatment regimen for use in a subject with neovascular age-related macular degeneration (nAMD), according to one or more embodiments.

[0015] [Diagram 5] 1 is a flowchart of a process for selecting a treatment regimen for use in a subject with neovascular age-related macular degeneration (nAMD), according to one or more embodiments.

[0016] [Figure 6] 1 is a table showing statistical results of different models according to one or more embodiments.

[0017] [Figure 7] FIG. 1 is a block diagram of a computer system in accordance with various embodiments.

[0018] In the accompanying drawings, similar components and / or features may have the same reference label. Furthermore, various components of the same type may be distinguished by following the reference label with a dash and a second label that distinguishes the similar components. If only a first reference label is used in this specification, the description is applicable to any one of the similar components having the same first reference label, regardless of the second reference label. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0019] I. Overview Neovascular age-related macular degeneration (nAMD) can be treated with anti-VEGF treatment, antibody treatment, another type of treatment, or a combination thereof. Two examples of anti-VEGF treatments are ranibizumab and aflibercept, which can be administered by intravitreal injection. One example of an antibody treatment is a monoclonal antibody treatment directed against vascular endothelial growth factor (VEGF), and angiopoietin 2 inhibitors. Typically, nAMD treatments are administered by injection (e.g., intravitreal injection) at a frequency ranging from about every 4 to about every 8 weeks. However, some patients may not require such frequent injections.

[0020] The frequency of treatments can generally be burdensome for patients and lead to reduced disease control in the real world. For example, after an initial stage of treatment, patients can be scheduled for regular monthly visits pro re nata (PRN) or for the required period. This PRN period can be, for example, 21 to 24 months, or some other monthly period. Visiting the clinic for monthly visits during the PRN period can be burdensome for patients who do not require frequent treatments. For example, if a patient requires no more than five injections during the entire PRN period, traveling for monthly visits can be overly cumbersome. Thus, patient compliance with visits can decrease over time, leading to reduced disease control.

[0021] Furthermore, the therapeutic response of neovascular age-related macular degeneration (nAMD) varies from subject to subject. For example, different subjects may respond differently to different injection frequencies associated with the same type of treatment. As an example, two subjects who receive the same treatment with the same injection frequency may experience different levels of visual acuity improvement. In some instances, two subjects may require different numbers of injections over time to achieve the same or similar visual acuity improvement. In other instances, two subjects may require two different types of treatment to achieve the same or similar visual acuity improvement.

[0022] Thus, the embodiments described herein recognize that it may be desirable to have a method that maximizes the improvement of vision in a subject with the least number or frequency of injections. The treatment regimen may identify a treatment (e.g., anti-VEGF therapy, monoclonal antibody therapy, etc.) and at least one of an administration frequency, an administration schedule, a monitoring schedule, or a combination thereof for the treatment. The administration frequency may be, for example, the frequency of injections (e.g., for intravitreal injections). The administration schedule may be, for example, the dosage per injection, which may remain constant or may change over time (e.g., a different dosage in an earlier injection compared to a later injection). The monitoring schedule may be, for example, a schedule of one or more monitoring visits or imaging sessions to evaluate the progress of the treatment.

[0023] The embodiments described herein provide a method and system for selecting a treatment regimen for a given subject with neovascular age-related macular degeneration (nAMD) to maximize the improvement of vision in the subject with the least number or frequency of injections. For example, multiple treatment regimens can be analyzed to determine the expected treatment response (or outcome) of each treatment regimen for the subject. The treatment regimen that produces the best response (or outcome) with the lowest treatment burden can be used as the selected treatment regimen. In some cases, the selected treatment regimen can be referred to as the optimal treatment regimen for the subject.

[0024] In one or more embodiments, various types of baseline data (e.g., baseline clinical data, baseline optical coherence tomography (OCT) image data, feature data extracted from baseline OCT image data, combinations thereof, etc.) are processed through one or more machine learning models to predict the treatment response to each of a plurality of treatment regimens. The prediction can be, for example, in the form of a treatment score. The treatment score can be, for example, a measurement of visual acuity at a future time point (e.g., 3 months, 6 months, 7 months, 8 months, 9 months, 10 months, 11 months, 12 months, 13 months, 15 months, 18 months, 20 months, 24 months, or some other number of months, days, or years after the treatment or the first administration of the treatment). In other examples, the treatment score can be some other type of measurement or metric indicative of the health of visual acuity.

[0025] In some cases, there may be no difference or the difference between the therapeutic response (or outcome) of two or more therapeutic regimens is minimal (e.g., less than a selected threshold).In these cases, the therapeutic regimen with the least treatment burden may be selected as the selected treatment.Treatment burden may be determined based on, for example, but not limited to, at least one of the following factors: the number and / or frequency of injections, the number or frequency of recommended or required follow-up visits, the possibility of one or more side effects or complications, the intensity of medication, or any other type of factor that makes the treatment more difficult to manage or maintain.

[0026] Recognizing and taking into account the importance and usefulness of the methodology and system that can bring about the above-mentioned improvements, the embodiments described herein allow for using machine learning models to predict the treatment regimen of a subject diagnosed with nAMD that minimizes the number or frequency of injections and maximizes the improvement of vision.Having the ability to personalize the treatment regimen selected for a subject based on the subject's baseline condition encourages to assist clinicians in making informed, personalized treatment decisions that can better benefit such subjects.Furthermore, making such decisions based on the treatment outcomes predicted by machine learning models can reduce the overall computing resources that would otherwise be required to analyze and identify the treatment regimen that is selected.

[0027] II. SYSTEM FOR SELECTING A TREATMENT REGIMEN FOR A SUBJECT WITH nAMD II.A. Overview FIG. 1 is a block diagram of a prediction system 100 according to various embodiments. The prediction system 100 may be used to predict a selected treatment regimen 101 that maximizes visual improvement while minimizing the burden of treatment for one or more subjects diagnosed with AMD, particularly nAMD. Such patient treatments may include, for example, but are not limited to, anti-VEGF treatments, antibody treatments, another type of treatment, or a combination thereof. Anti-VEGF treatments may include, for example, ranibizumab, which may be administered by intravitreal injection. Antibody treatments may include, for example, but are not limited to, monoclonal antibody treatments directed against vascular endothelial growth factor (VEGF) and angiopoietin 2 inhibitors. In one or more embodiments, the antibody treatments include faricimab.

[0028] As mentioned above, the treatment regimen may identify a treatment (e.g., anti-VEGF therapy, monoclonal antibody therapy, etc.) and at least one of the following: administration frequency, administration schedule, monitoring schedule, or combination thereof for the treatment. The treatment regimen may also be referred to as a treatment arm in some cases. The administration frequency may be, for example, the frequency of injection (e.g., for intravitreal injection). The administration schedule may be, for example, the dosage per injection, which may remain constant or may change over time (e.g., the dosage in the previous injection is different compared to the subsequent injection). The monitoring schedule may be, for example, the schedule of one or more monitoring visits or imaging sessions to evaluate the progress of the treatment.

[0029] In one or more embodiments, the selected therapeutic regimen 101 may be a therapeutic regimen that maximizes the improvement of the subject's vision while minimizing the burden of treatment. The burden of treatment may be determined, for example, but not limited to, based on at least one of the following: the number and / or frequency of injections, the number or frequency of recommended or required follow-up visits, the likelihood of one or more side effects or complications, the intensity of the medication, or any other type of factor that makes the treatment more difficult to manage or maintain. As an example, the selected therapeutic regimen 101 may be a therapeutic regimen that provides the best therapeutic response (or outcome) while having the fewest number of injections or the lowest frequency of injections. In some cases, the selected therapeutic regimen 101 may be referred to as an optimal therapeutic regimen.

[0030] The prediction system 100 includes a computing platform 102, a data storage 104, and a display system 106. The computing platform 102 may take a variety of forms. In one or more embodiments, the computing platform 102 includes a single computer (or computer system) or multiple computers in communication with each other. In other examples, the computing platform 102 takes the form of a cloud computing platform. In some examples, the computing platform 102 takes the form of a mobile computing platform (e.g., a smartphone, a tablet, a smartwatch, etc.).

[0031] The data storage 104 and the display system 106 each communicate with the computing platform 102. In some examples, the data storage 104, the display system 106, or both may be considered part of the computing platform 102 or may be otherwise integrated with the computing platform 102. Thus, in some examples, the computing platform 102, the data storage 104, and the display system 106 may be separate components that communicate with each other, while in other examples, some combination of these components may be integrated together. The data storage 104 may include, for example, but is not limited to, a database, a spreadsheet, a server, cloud-based storage, or a combination thereof.

[0032] The predictive system 100 includes a data analyzer 108, which may be implemented using hardware, software, firmware, or a combination thereof. In one or more embodiments, the data analyzer 108 is implemented on the computing platform 102.

[0033] The data analyzer 108 includes an outcome prediction system 110 and a regimen predictor 112, each of which may be implemented using hardware, software, firmware, or a combination thereof. In some embodiments, the outcome prediction system 110 and the regimen predictor 112 may be implemented as separate modules. In other embodiments, at least a portion of the regimen predictor 112 may be implemented as part of the outcome prediction system 110.

[0034] The outcome prediction system 110 uses the baseline data 114 to generate (or predict) a plurality of treatment scores 116 for a corresponding plurality of treatment regimens 118. In some embodiments, the baseline data 114 may be received, at least in part, from a source external to the prediction system 100 via one or more communications links (e.g., wired communications links, wireless communications links, optical communications links, etc.). In one or more embodiments, the baseline data 114 is retrieved, at least in part, from the data storage 104.

[0035] The therapeutic regimen 118 may include two therapeutic regimens, three therapeutic regimens, four therapeutic regimens, or some other therapeutic regimen that may be administered to a subject. The therapeutic regimen 118 may differ with respect to treatment type, administration schedule, drug strength or concentration, number of injections, injection frequency, monitoring schedule, or a combination thereof. In this way, two different therapeutic regimens may correspond to the same treatment, but may also differ by other factors (e.g., dosage and / or injection frequency).

[0036] The treatment scores 116 may include a corresponding treatment score for each of the treatment regimens 118. Each of the treatment scores 116 may be a score indicative of a treatment response or outcome. As an example, a higher treatment score indicates a better response (e.g., greater improvement to vision) and a lower treatment score indicates a poorer response. In another example, a lower treatment score indicates a better response (e.g., greater improvement to vision) and a higher treatment score indicates a poorer response.

[0037] The treatment score may indicate, for example, but is not limited to, at least one of a predicted vision measurement (e.g., predicted best corrected visual acuity (BCVA)), a predicted vision change (e.g., predicted change in BCVA), a predicted CST, a predicted decline in CST, or some other type of response metric for a subject undergoing treatment. Treatment scores may be generated for a selected time point after treatment, for example, but is not limited to, 6 months, 9 months, 12 months, 24 months, or some other time after treatment.

[0038] In one or more embodiments, the treatment score 116 can be a predicted visual acuity measurement (e.g., BCVA) for the subject at a future time point after treatment or after the first administration of treatment. This future time point can be, for example, 3 months, 6 months, 7 months, 8 months, 9 months, 10 months, 11 months, 12 months, 13 months, 15 months, 18 months, 20 months, 24 months, or some other number of months, days, or years. In one or more embodiments, the treatment score 116 can be a metric that represents or is generated based on a predicted visual acuity measurement for the subject at a future time point.

[0039] The baseline data 114 may include data from a baseline time point. The baseline time point may be, for example, a time point prior to treatment or a time point contemporaneous with the first treatment administration (e.g., day 1 of treatment). In other embodiments, data from a time point during treatment (e.g., x(e.g., 1 week, 2 weeks, 4 weeks, 1 month, 2 months, 3 months after the administration of the 3rd treatment). Examples of different types of baseline data are described in more detail below in Section II.B.

[0040] The outcome prediction system 110 may be implemented in different ways to process the baseline data 114 and generate the treatment score 116. Examples of implementations of the outcome prediction system 110 are described in more detail below in Section II.B.

[0041] The regimen predictor 112 may be used to identify a selected therapeutic regimen 101 for use with a subject based on a plurality of treatment scores 116. In one or more embodiments, the regimen predictor 112 uses a set of selected criteria 119 to identify one of the therapeutic regimens 118 as the selected therapeutic regimen 101. The set of selected criteria 119 may include a variety of criteria. For example, the set of selected criteria 119 may include a score criterion, a set of treatment burden criteria, or both. The set of treatment burden criteria may include, for example, but is not limited to, at least one of the following: least number of injections, lowest frequency of injections, lowest dosage, lowest drug strength, reduced degree of monitoring, or reduced side effects.

[0042] In one or more embodiments, the score criterion is a treatment score criterion. For example, the score criterion can be the highest treatment score or the lowest treatment score. In some embodiments, the score criterion can be for a subset of the treatment scores 116 (e.g., a subset of the treatment scores 116 that fall within a selected range, a subset of the treatment scores above a selected threshold, a subset of the treatment scores below a selected threshold, or some other subset). In these examples, the score criterion can be the highest treatment score or the lowest treatment score of the subset.

[0043] In other embodiments, the regimen predictor 112 may calculate new scores based on the treatment scores 116. These new scores may in some cases be different types of indicators of treatment response or outcome. In these examples, the score criteria may be for these calculated scores. For example, the score criteria may be the highest calculated score or the lowest calculated score. In some cases, the score criteria may be for a subset of the calculated scores (e.g., a subset of the calculated scores that fall within a selected range, a subset of the calculated scores that are above a selected threshold, a subset of the calculated scores that are below a selected threshold, or some other subset).

[0044] In one or more embodiments, the regimen predictor 112 may identify the treatment regimen of the treatment regimen 118 that most closely meets the set of selected criteria 119 as the selected treatment regimen 101 for use with the subject. As an example, the regimen predictor 112 may first determine whether a single treatment regimen or multiple treatment regimens of the treatment regimen 118 meets the criteria of the score. If the single treatment regimen meets the criteria of the score, the regimen predictor 112 identifies the single treatment regimen as the selected treatment regimen 101, thereby predicting that the selected treatment regimen 101 will maximize vision improvement while minimizing the treatment burden of the subject. However, if multiple treatment regimens meet the criteria of the score, the regimen predictor 112 identifies the treatment regimen that meets the set of the treatment burden criteria of the subject as the selected treatment regimen 101.

[0045] For example, if the score criterion is the highest treatment score, the regimen predictor 112 may select the treatment regimen having the highest treatment score as the selected treatment regimen 101. However, in certain examples, two or more of the treatment regimens 118 may be tied. For example, two or more treatment regimens may be tied by both having the highest treatment score. As another example, two or more treatment regimens may be tied by having the two highest treatment scores with the difference between the two treatment scores being below a selected threshold.

[0046] If so, the regimen predictor 112 may select the therapeutic regimen that meets a set of treatment burden criteria (e.g., that which imposes the least treatment burden on the subject) as the selected therapeutic regimen 101. In one or more embodiments, the regimen predictor 112 uses the regimen data 120 to identify which tied therapeutic regimen meets a set of treatment burden criteria (e.g., imposes the least burden on the subject).

[0047] Regimen data 120 may specify details of each of the therapeutic regimens 118. Regimen data 120 may specify, for example, without limitation, a type of treatment and at least one of an administration frequency, an administration schedule, a monitoring schedule, or a combination thereof for the treatment. In some cases, regimen data 120 may further specify other information regarding each of the therapeutic regimens 118. For example, regimen data 120 may specify information regarding side effects or complications, dosage intensity, and / or information regarding the burden imposed on a given subject by a given therapeutic regimen.

[0048] The data analyzer 108 may use the selected therapeutic regimen 101 and possibly a plurality of treatment scores 116, regimen data 120, or both to form a final output 122. The final output 122 may, for example, identify the selected therapeutic regimen 101. In some cases, the final output 122 may include a portion of the information in the regimen data 120 that is related to the selected therapeutic regimen 101. In some examples, the final output 122 includes treatment scores 116 for the entire therapeutic regimen 118. These treatment scores 116 may be presented in chart format, graph, table, ranked list, or combinations thereof. In other examples, the final output 122 includes only the treatment scores corresponding to the selected therapeutic regimen 101. In yet other examples, the final output 122 includes only a selected number of the highest (or alternatively, lowest) treatment scores. For example, the final output 122 may include the top two treatment scores.

[0049] In one or more embodiments, the final output 122 includes other information generated based on the selected treatment regimen 101. For example, the prediction of the selected treatment regimen 101 may be used to include or exclude a subject from a clinical trial. For example, if a subject is assigned to a clinical trial of one of multiple possible treatment regimens, the data analyzer 108 may be used to generate a final output 122 indicating which clinical trial is expected to be most successful for the subject in terms of the lowest treatment burden. In this manner, the final output 122 may be used to improve overall treatment management as well as the enrichment of the clinical trial population.

[0050] In one or more embodiments, at least a portion of the final output 122, or a graphical representation of at least a portion of the final output 122, is displayed on the display system 106. In some embodiments, at least a portion of the final output 122, or a graphical representation of at least a portion of the final output 122, is transmitted to a remote device 124 (e.g., a mobile device, a laptop, a server, the cloud, etc.).

[0051] II.B. Exemplary Outcome Predictors FIG. 2 is a block diagram of the outcome prediction system 110 from FIG. 1, according to one or more embodiments. The outcome prediction system 110 will be described with continued reference to the elements shown in FIG. 1. The outcome prediction system 110 is used to predict a plurality of treatment scores 116 for a corresponding plurality of treatment regimens 118, as described above in Section II.A. The outcome prediction system 110 may include an outcome predictor 200. The outcome predictor 200 may include, for example, a set of predictor models 204. The set of predictor models 204 may be one or more predictor models. In one or more embodiments, each predictor model in the set of predictor models 204 includes one or more machine learning models. In this manner, the set of predictor models 204 may be composed of a set of machine learning models 205. When the set of predictor models 204 includes multiple predictor models, these predictor models may be implemented in the same or similar manner. In some cases, each of these predictor models may be implemented to use the same machine learning model or the same machine learning architecture.

[0052] In one or more embodiments, the outcome predictor 200 may be used to predict a treatment score 116 for a treatment regimen 118 described in FIG. The treatment score 203 may be one example of a treatment score 116. The treatment score 116 may be or may indicate at least one of a predicted vision measurement (e.g., predicted best corrected visual acuity (BCVA)), a predicted vision change (e.g., predicted change in BCVA), a predicted CST, a predicted decline in CST, or some other type of response metric for a subject undergoing treatment. The treatment score 203 may be generated for a selected time point after treatment, for example, but not limited to, 6 months, 9 months, 12 months, 24 months, or some other time after treatment.

[0053] The outcome prediction system 110 may receive, and possibly also generate, baseline data 114. The baseline data 114 may be used to generate a plurality of predictor inputs 206 that are processed by the outcome predictor 200 to generate a treatment score 116. A plurality of predictor inputs 206 may be generated for treatment regimens 118 such that the plurality of predictor inputs 206 includes one predictor input for each of the treatment regimens 118.

[0054] In one or more embodiments, the baseline data 114 includes, but is not limited to, clinical data 208, imaging data 210, or a combination thereof for a baseline time point. The clinical data 208 may include, for example, but is not limited to, at least one of baseline demographic data, baseline visual acuity measurements, baseline central subfield thickness (CST) measurements, baseline low-light defect (LLD), or some other type of baseline measurement. The baseline attribute data may include, for example, but is not limited to, at least one of age, sex, or another type of attribute metric. The baseline visual acuity measurement may be, for example, a best corrected visual acuity (BCVA) measurement. The baseline CST measurement may be, for example, in micrometers. The baseline LLD may be the difference between the baseline BCVA measurement and the baseline low-light visual acuity (LLVA) measurement.

[0055] The imaging data 210 may include one or more types of imaging data for a baseline time point. For example, the imaging data 210 may include optical coherence tomography (OCT) image data, color fundus image data, infrared image data, near-infrared image data, fundus autofluorescence image data, fluorescein angiography image data, another type of three-dimensional imaging data, or a combination thereof. In one or more embodiments, the imaging data 210 includes three-dimensional OCT imaging data, data extracted from an OCT image (e.g., an OCT en face image), tabular data extracted from an OCT image, some other form of imaging data, or a combination thereof. The OCT imaging data may include, for example, a spectral domain OCT (SD-OCT) B-scan that together form an OCT volume scan of the subject.

[0056] In some embodiments, the outcome prediction system 110 may include other models that may be used to generate the baseline data 114 for use in forming the plurality of predictor inputs 206. For example, the outcome prediction system 110 may include a segmentation model 212, a feature extraction module 214, or both.

[0057] The segmentation model 212 may be used to perform an automatic segmentation of the imaging data 210 (e.g., of the OCT imaging data) to form the segmented image data 216. The segmentation model 212 may be implemented to use a deep learning model (e.g., one or more neural networks). For example, the segmentation model 212 may be implemented to use one or more convolutional neural networks. In one or more embodiments, the segmentation model 212 includes a U-net. In some embodiments, at least a portion of the segmented image data 216 is used to form the multiple predictor inputs 206 for the outcome predictor 200. In some embodiments, at least a portion of the segmented image data 216 may be transmitted to the feature extraction module 214 for further processing. For example, in some cases, each of the B-scans (e.g., 128 B-scans) forming the OCT volume scan of the imaging data 210 may be segmented to form the segmented image data 216 (e.g., 128 segmented images). However, in certain cases, only a portion of the segmented images (e.g., 20, 30, 40, 50, etc.) are sent to the feature extraction module 214 for processing or used to form the multiple predictor inputs 206.

[0058] Segmentation via segmentation model 212 includes detection and identification of one or more retinal (e.g., retina-related) elements of the retinal image. The segmented image of segmented image data 216 identifies one or more retinal (e.g., retina-related) elements of the segmented image using one or more graphical indicators. The segmented image may be a representation of an OCT image identifying one or more retinal elements, or may be an OCT image in which one or more retinal elements have been identified.

[0059] For example, one or more color indicators, shape indicators, pattern indicators, shading indicators, lines, curves, markers, labels, tags, text features, other types of graphical indicators, or combinations thereof may be used to identify portions of the image (e.g., pixel by pixel) identified as retinal elements. As one specific example, a group of pixels may be identified as acquiring a particular retinal fluid (e.g., IRF or SRF). The segmented image may identify this group of pixels using a color indicator. For example, each pixel of the group of pixels may be assigned a color unique to a particular retinal fluid, thereby assigning each pixel to the particular retinal fluid. As another example, the segmented image may identify the group of pixels by applying a patterned region or shape (continuous or discontinuous) over the group of pixels.

[0060] The retinal element may be composed of at least one of a retinal layer element or a retinal pathology element. The detection and identification of one or more retinal layer elements may be referred to as segmenting a layer element (or a retinal layer element). The detection and identification of one or more retinal pathology elements may be referred to as segmenting a pathology element (or a retinal pathology element).

[0061] The retinal layer element may be, for example, a retinal layer or a boundary associated with a retinal layer. Examples of retinal layers include, but are not limited to, the inner limiting membrane (ILM) layer, the retinal nerve fiber layer, the ganglion cell layer, the inner plexiform layer, the inner nuclear layer, the outer plexiform layer, the outer nuclear layer, the outer limiting membrane (ELM) layer, the photoreceptor layer, the retinal pigment epithelium (RPE) layer, the layer of RPE detachment, the Bruch's membrane (BM) layer, the choriocapillaris layer, the choroidal stromal layer, the ellipsoid zone (EZ), and other types of retinal layers. In some cases, a retinal layer may be composed of one or more layers. As an example, a retinal layer may be the interface between the outer plexiform layer and the fiber layer of Henle (OPL-HFL). The boundary associated with a retinal layer may be, for example, the inner boundary of a retinal layer, the outer boundary of a retinal layer, a boundary associated with a pathological feature of a retinal layer (e.g., the inner or outer boundary of a retinal layer detachment), or some other type of boundary. For example, the boundary may be an inner boundary of a separation layer of the RPE (IB-RPE), an outer boundary of a separation layer of the RPE (OB-RPE), or another type of boundary.

[0062] Retinal pathological elements may include, for example, fluid (e.g., fluid pockets), cells, solid material, or combinations thereof, indicative of retinal pathology (e.g., a disease or condition such as AMD or diabetic macular edema). For example, the presence of certain retinal fluid may be a sign of nAMD. Examples of retinal pathological elements include, but are not limited to, intraretinal fluid (IRF), subretinal fluid (SRF), fluid associated with pigment epithelial detachment (PED), hyperreflective material (HRM), subretinal hyperreflective material (SHRM), intraretinal hyperreflective material (IHRM), hyperreflective foci (HRF), retinal fluid pockets, drusen, development of fibrosis, and disruption. In some cases, retinal pathological elements may be disruption (e.g., discontinuity, delamination, loss, etc.) of a retinal layer or zone. For example, the disruption may be disruption of the ellipsoid region, ELM, RPE, or another layer or region. The disruption may represent damage or loss of cells (e.g., photoreceptors) in the area of ​​the disruption. In some instances, the retinal pathology may be clear IRF, cloudy IRF, clear SRF, cloudy SRF, any other type of clear retinal fluid, any other type of cloudy retinal fluid, or a combination thereof.

[0063] The feature extraction module 214 may be used to extract retinal feature data 218 from the segmented image data 216 generated by the segmentation model 212 or aggregate retinal feature data 218 from data received directly from the segmentation model 212. In one or more embodiments, at least a portion of the retinal feature data 218 may be used to form the plurality of predictor inputs 206. The retinal feature data 218 may include, for example, but not limited to, values ​​of any number or combination of features (e.g., quantitative retinal features). These features may include features associated with pathology, volumetric features associated with layers, thickness features associated with layers, or combinations thereof. Examples of features include, but are not limited to, maximum retinal layer thickness, minimum retinal layer thickness, average retinal layer thickness, maximum height of a boundary associated with a retinal layer, volume of a retinal fluid pocket, length of a fluid pocket, width of a fluid pocket, number of retinal fluid pockets, and a plurality of hyperreflective foci. Thus, at least some of the features may be volumetric features. For example, feature data may be derived for each selected OCT image (e.g., a single OCT B-scan) and then combined to form a volume-wide value. In one or more embodiments, between 1 and 200 features may be included in the retinal feature data 218.

[0064] In this manner, at least a portion of the clinical data 208, at least a portion of the imaging data 210, at least a portion of the segmented image data 216, at least a portion of the retinal feature data 218, or a combination thereof may be used to form each predictor input of the plurality of predictor inputs 206 that is sent to the outcome predictor 200 for processing. This information may be the same for each of the plurality of predictor inputs 206.

[0065] Additionally, each of the plurality of predictor inputs 206 may further include at least a portion of the regimen data 120 of the corresponding treatment regimen. For example, if the set of predictor models 204 includes a single predictor model (e.g., may include one or more machine learning models), each predictor input of the plurality of predictor inputs 206 may be independently fed into the single predictor model. Each predictor input includes a portion of the regimen data 120 of its corresponding treatment regimen. In this manner, each of the plurality of predictor inputs 206 is independently processed through the single predictor model (e.g., via a different forward pass) to generate the treatment score 116.

[0066] In other examples, where the set of predictor models 204 includes multiple predictor models, each corresponding to a different one of the treatment regimens 118, each of the multiple predictor models 206 may be sent to its corresponding predictor model to collectively generate the treatment score 116. In this manner, each of the multiple predictor models is specifically tuned to generate a treatment score for the corresponding treatment regimen.

[0067] The outcome predictor 202 may be implemented in a variety of different ways. In one or more embodiments, the outcome predictor 202 may be implemented using any number or combination of artificial intelligence or machine learning models. The outcome predictor 202 may be composed of multiple models and / or algorithms. For example, each predictor model in the set of predictor models 204 may include any number or combination of regression models, linear models, random forest models, XGBoost algorithms, deep learning models (e.g., neural networks), and / or other types of machine learning models.

[0068] In one or more embodiments, the outcome predictor 202 (e.g., one of the set of predictor models 204) includes a deep learning model and a symbolic model (e.g., a classical machine learning model). In one or more embodiments, the deep learning model includes one or more neural networks, at least one of which is a deep learning neural network (or deep neural network (DNN)). In one or more embodiments, the symbolic model includes one or more models that use symbolic machine learning. The symbolic model may include, for example, but is not limited to, at least one of a linear model, a random forest model, an extreme gradient boosting (XGBoost) algorithm, a support vector machine (SVM) model, or another type of model or algorithm that uses symbolic learning.

[0069] In various embodiments, the outcome predictor 202 may use a model stacking approach to generate the treatment score 203. In the model stacking approach, for example, a first machine learning model (e.g., a deep learning model) may be used to generate an intermediate treatment score that is sent as input to a second machine learning model (e.g., a symbolic model). The intermediate treatment score, along with other data (e.g., baseline clinical data), may be sent as input to the second machine learning model for processing. The second machine learning model then generates the treatment score 203.

[0070] In various embodiments, the outcome predictor 202 may use a model averaging approach to generate the treatment score 203. In a model averaging approach, at least two different machine learning models may generate an intermediate treatment score using multiple predictor inputs 206. These intermediate treatment scores may then be averaged (e.g., via equal weighting) to generate the treatment score 203.

[0071] Together, the outcome predictor 200 generates a treatment score 116. The treatment score 116 can then be transmitted to the regimen predictor 112 described above with respect to FIG. 1 to predict a selected therapeutic regimen 101 to use for the subject.

[0072] II.C. Exemplary Workflow for Predicting Selected Treatment Regimen FIG. 3 is a schematic diagram of an exemplary workflow for predicting a selected treatment regimen. The workflow 300 is described with continued reference to the elements of FIG. 1 and FIG. 2. The workflow 300 illustrates how the baseline data 114 may be processed using machine learning to predict the selected treatment regimen 101. The predictor model 302 may be an example of a predictor model in the set of predictor models 204 in the outcome predictor 200 of FIG. 2. The predictor model 302 may be used on multiple occasions (e.g., multiple cycles of treatment) for multiple treatment regimens. For example, for three different treatment regimens, the predictor model 302 may be run at three different times using three different inputs. The predictor model 302 may be implemented using a machine learning model (e.g., a linear regression model, SVM, XGBoost algorithm, random forest, etc.).

[0073] The baseline data 114 may be used as an input for each instance of the predictor model 302. In FIG. 3, the baseline data 114 includes a baseline visual acuity measurement, a baseline CST measurement, a baseline age, and a gender. The first treatment regimen data 306 may be used as an input for the first instance of the predictor model 302. The second treatment regimen data 307 may be used as an input for the second instance of the predictor model 302. The third treatment regimen data 310 may be used as an input for the third instance of the predictor model 302. The first treatment regimen data 306, the second treatment regimen data 307, and the third treatment regimen data 310 may be different portions of the regimen data 120 described with respect to FIG. 1 and FIG. 2, each identifying a different corresponding treatment regimen (e.g., a different one of the treatment regimens 118 of FIG. 1).

[0074] During each case or execution cycle (or forward pass), the predictor model 302 processes the received input. The predictor model 302 generates a treatment score for the case's corresponding specified treatment regimen. For example, in a first case, the predictor model 302 generates a treatment score 312, in a second case, the predictor model 302 generates a treatment score 314, and in a third case, the predictor model 302 generates a treatment score 316.

[0075] These treatment scores are processed by the regimen predictor 112. The regimen predictor 112 uses a set of selected criteria, which may include a score criterion, to identify the therapeutic regimen 101 to be selected. For example, the regimen predictor 112 may use a maximum function that identifies the maximum of the inputs provided to the maximum function. In other words, the regimen predictor 112 identifies a treatment score that meets a score criterion that is the maximum treatment score.

[0076] III. Exemplary Methods for Predicting Selected Treatment Regimen 4 is a flow chart of a process for predicting a selected treatment regimen according to one or more embodiments. In one or more embodiments, process 400 may be implemented using prediction system 100 described in FIG. 1 and / or outcome prediction system 110 described in FIGS. 1-2. Process 400 includes various steps and may be described with continued reference to FIGS. 1-2. One or more steps not explicitly shown in FIG. 4 may be included before, after, between, or as part of a step of process 400. In some embodiments, process 400 may begin with step 402.

[0077] Step 402 includes receiving baseline data of a subject diagnosed with neovascular age-related macular degeneration (nAMD). The baseline data may be, for example, baseline data 114 of FIGS. 1-2. For example, the baseline data may include clinical data, imaging data, segmented image data, retinal feature data, or combinations thereof for a baseline time point. The clinical data may include, for example, at least one of baseline age, sex, baseline CST, baseline visual acuity measurement (e.g., baseline BCVA), or some other type of baseline factor.

[0078] Step 404 includes forming a plurality of predictor inputs for the outcome predictor using the baseline data and the regimen data for the plurality of treatment regimens. The plurality of predictor inputs includes a different predictor input for each of the plurality of treatment regimens. The outcome predictor includes at least one machine learning model. For example, the outcome predictor may include one or more predictor models, each of which may be comprised of at least one of a linear regression model, a random forest (RF) model, a gradient boosting machine (GBM) model, an extreme gradient boosting (XGBoost), or a support vector machine (SVM) model. In other embodiments, the outcome predictor may include a deep learning model (e.g., one or more neural networks). The outcome predictor may be, for example, the outcome predictor 200 of FIG. 2.

[0079] The plurality of predictor inputs may be, for example, but not limited to, the plurality of predictor inputs 206 of FIG. 2. The regimen data may be, for example, the regimen data 120 of FIGS. 1-2. The regimen data of a corresponding one of the plurality of therapeutic regimens identifies a treatment and at least one of an administration frequency, an administration schedule, or a monitoring schedule of the treatment. Further, each of the plurality of predictor inputs 206 corresponds to a different one of the plurality of therapeutic regimens by including therapeutic regimen data identifying that particular therapeutic regimen.

[0080] Predictor inputs for the corresponding treatment regimen may also include, for example, but not limited to, clinical data and retinal feature data. The retinal feature data may be included in the baseline data received in step 402. In another example, the retinal feature data may be extracted from segmented image data included in the baseline data. In yet another example, the retinal feature data may be extracted from segmented image data generated based on imaging data (e.g., a baseline OCT volume scan) included in the baseline data received in step 402.

[0081] Step 406 includes generating a plurality of treatment scores for a plurality of treatment regimens via the outcome predictor. Each of the treatment scores may be, for example, a predicted visual acuity measurement (e.g., predicted BCVA), a predicted visual acuity change (e.g., predicted change in BCVA), a predicted CST, a predicted decline in CST, or some other type of response metric for a subject undergoing treatment. The treatment scores may be generated for a selected time point after treatment, for example, but not limited to, 6 months, 9 months, 12 months, 24 months, or some other time after treatment. In one example, the treatment score is predicted BCVA at 9 months.

[0082] Step 408 includes selecting one of the multiple treatment regimens as the subject's selected treatment regimen based on the multiple treatment scores. The selected treatment regimen maximizes vision improvement while minimizing treatment burden (e.g., number of injections, injection frequency). In some embodiments, step 408 includes determining whether a single treatment regimen of the multiple treatment regimens, or multiple treatment regimens of the multiple treatment regimens, meets a score criterion. The score criterion may be, for example, the highest (maximum) treatment score, the lowest (minimum) treatment score, the highest calculated score (e.g., a score calculated or otherwise derived from the treatment scores), the lowest calculated score, or some other type of treatment score. In some cases, the treatment scores or the calculated scores based on the treatment scores may include letters or other text. The score criterion may be selected to identify the best predicted treatment outcome or response based on the treatment scores or the calculated scores.

[0083] Multi-treatment regimen can meet the criteria of score by being tied.For example, two treatment scores can be tied by sharing the same value.In other cases, scores can be tied by having a difference less than a selected threshold.For example, but not limited to, a numerical difference of 1 or less can be considered as a tie.

[0084] If a single treatment regimen meets the criteria of the score, step 408 may include selecting the single treatment regimen as the selected treatment regimen for the subject. However, if multiple treatment regimens meet the criteria of the score, step 408 may include identifying a treatment regimen among the multiple treatment regimens that meets a set of treatment burden criteria for the subject as the selected treatment regimen. The set of treatment burden criteria may include, for example, but is not limited to, at least one of the following: least number of injections, lowest injection frequency, lowest dosage, lowest drug strength, reduced degree of monitoring, or reduced side effects.

[0085] 5 is a flow chart of a process for predicting a selected treatment regimen according to one or more embodiments. In one or more embodiments, the process 500 may be implemented using the prediction system 100 described in FIG. 1 and / or the outcome prediction system 110 described in FIGS. 1-2. The process 500 includes various steps and may be described with continued reference to FIGS. 1-2. One or more steps not explicitly shown in FIG. 5 may be included before, after, between, or as part of a step of the process 500. In some embodiments, the process 500 may begin with step 502.

[0086] Step 502 includes receiving clinical and imaging data for a subject diagnosed with neovascular age-related macular degeneration (nAMD) for a baseline time point. The clinical and imaging data may be, for example, clinical data 208 and imaging data 210 of FIG. 2.

[0087] Step 504 includes generating retinal feature data using the imaging data, the retinal feature data including at least one of a pathology-related feature, a layer-related volumetric feature, or a layer-related thickness. The retinal feature data may be, for example, the retinal feature data 218 of FIG. 2.

[0088] Step 506 includes generating a plurality of treatment scores for a plurality of treatment regimens via an outcome predictor that includes at least one machine learning model. In one or more embodiments, each of the machine learning models may be implemented using a regression model, an XGBoost, an SVM model, a random forest, and / or another type of machine learning model. Each of the treatment scores may be, for example, a predicted visual acuity measurement (e.g., predicted BCVA), a predicted visual acuity change (e.g., predicted change in BCVA), a predicted CST, a predicted decline in CST, or some other type of response metric for a subject undergoing treatment. The treatment scores may be generated for a selected time point after treatment, for example, but not limited to, 6 months, 9 months, 12 months, 24 months, or some other time after treatment.

[0089] Step 508 includes selecting one of the plurality of therapeutic regimens as the selected therapeutic regimen for the subject based on the plurality of treatment scores. Step 508 may be performed in a manner similar to step 408 of process 400 described above with respect to FIG.

[0090] IV. Illustrative Study Results In one study, the dataset consisted of eyes of 324 subjects treated with ranibizumab 0.5 mg Q4W (i.e., every 4 weeks). These subjects were treatment naive prior to treatment with ranibizumab. Subjects' baseline BCVA ranged from 20 / 40 to 20 / 320 (i.e., Early Treatment Diabetic Retinopathy Severity Scale (ETDRS) letter 73 to 24). All subjects were aged 50 years or older.

[0091] The system was trained to predict a selected treatment regimen (e.g., optimal treatment regimen) for use in a subject based on benchmark features (e.g., age, sex, baseline BCVA, baseline CST). It was also trained to identify a selected treatment regimen based on the benchmark features and retinal feature data obtained from imaging data (e.g., from segmented images generated from SD-OCT images).

[0092] A linear model, a random forest model, an XGBoost model, and a support vector machine model were each developed to predict treatment scores (e.g., predicted BCVA at month 9) for both benchmark features and benchmark features plus retinal features. The regimen with the highest treatment score was selected as the treatment regimen of choice. The performance of the models was evaluated using R 2 The scores (i.e., coefficient of determination) were used for evaluation.

[0093] 6 is a table 600 showing statistical results of different models according to one or more embodiments. Table 600 shows that using a linear model, a random forest model, an XGBoost model, and a support vector machine model to calculate a treatment score based on both benchmark features and image-derived retinal features performed well in this particular study, with XGBoost generally performing the best.

[0094] V. Computer Implemented Systems FIG. 7 is a block diagram illustrating an example of a computer system according to various embodiments. The computer system 700 may be an example of one implementation of the computing platform 102 described above in FIG. 1. In one or more examples, the computer system 700 may include a bus 702 or other communication mechanism for communicating information and a processor 704 coupled with the bus 702 for processing information. In various embodiments, the computer system 700 may also include memory, which may be a random access memory (RAM) 706 or other dynamic storage device, coupled to the bus 702 for determining instructions to be executed by the processor 704. The memory may also be used to store temporary variables or other intermediate information during execution of instructions to be executed by the processor 704. In various embodiments, the computer system 700 may further include a read only memory (ROM) 708 or other static storage device coupled to the bus 702 for storing static information and instructions for the processor 704. A storage device 710, such as a magnetic disk or optical disk, may be provided and coupled to the bus 702 for storing information and instructions.

[0095] In various embodiments, the computer system 700 can be coupled via the bus 702 to a display 712, such as a cathode ray tube (CRT) or liquid crystal display (LCD), for displaying information to a computer user. An input device 714, including alphanumeric and other keys, can be coupled to the bus 702 for communicating information and command selections to the processor 704. Another type of user input device is a cursor control 716, such as a mouse, joystick, trackball, gesture input device, gaze-based input device, or cursor direction keys, for communicating directional information and command selections to the processor 704 and for controlling cursor movement on the display 712. This input device 716 typically has two degrees of freedom in two axes, a first axis (e.g., x) and a second axis (e.g., y), that allow the device to specify a position in a plane. However, it should be understood that input devices 716 that allow three-dimensional (e.g., x, y, and z) cursor movement are also contemplated herein.

[0096] Consistent with a particular implementation of the present teachings, results can be provided by computer system 700 in response to processor 704 executing one or more sequences of one or more instructions contained in RAM 706. Such instructions can be read into RAM 706 from another computer-readable medium or computer-readable storage medium, such as storage device 710. Execution of the sequences of instructions contained in RAM 706 can cause processor 704 to perform the processes described herein. Alternatively, hardwired circuitry can be used in place of or in combination with software instructions to implement the present teachings. Thus, implementation of the present teachings is not limited to any specific combination of hardware circuitry and software.

[0097] As used herein, the terms "computer-readable medium" (e.g., data store, data storage, data storage device, etc.) or "computer-readable storage medium" refer to any medium that participates in providing instructions to processor 704 for execution. Such media may take many forms, including but not limited to, non-volatile media, volatile media, and transmission media. Examples of non-volatile media may include, but are not limited to, optical, solid-state, and magnetic disks, such as storage device 710. Examples of volatile media may include, but are not limited to, dynamic memory, such as RAM 706. Examples of transmission media may include, but are not limited to, coaxial cables, copper wire, and fiber optics, including the wires that comprise bus 702.

[0098] Common forms of computer readable media include, for example, a floppy disk, a flexible disk, a hard disk, a magnetic tape or any other magnetic medium, a CD-ROM, any other optical medium, punch cards, paper tape, any other physical medium with a pattern of holes, RAM, PROM, and EPROM, flash EPROM, any other memory chip or cartridge, or any other tangible medium from which a computer can read.

[0099] In addition to computer-readable media, instructions or data can be provided as signals on a transmission medium included in a communication device or system to provide a sequence of one or more instructions to the processor 704 of the computer system 700 for execution. For example, a communication device may include a transceiver having signals indicative of instructions and data. The instructions and data are configured to cause one or more processors to implement the functions outlined in the disclosure herein. Representative examples of data communication transmission connections include, but are not limited to, a telephone modem connection, a wide area network (WAN), a local area network (LAN), an infrared data connection, an NFC connection, an optical communication connection, and the like.

[0100] It should be understood that the methods, flowcharts, diagrams, and accompanying disclosure described herein can be implemented using computer system 700 as a stand-alone device or on a distributed network of shared computer processing resources, such as a cloud computing network.

[0101] The methodologies described herein may be implemented by various means depending on the application. For example, these methodologies may be implemented in hardware, firmware, software, or any combination thereof. In the case of a hardware implementation, the processing unit may be implemented in one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), processors, controllers, microcontrollers, microprocessors, electronic devices, other electronic units designed to perform the functions described herein, or combinations thereof.

[0102] In various embodiments, the methods of the present teachings may be implemented as firmware and / or software programs and applications written in conventional programming languages ​​such as C, C++, Python, etc. When implemented as firmware and / or software, the embodiments described herein may be implemented in a non-transitory computer-readable medium having stored thereon a program for causing a computer to execute the above-mentioned methods. It should be understood that the various engines described herein may be provided on a computer system such as computer system 700, whereby processor 704 performs the analysis and decisions provided by these engines according to instructions provided by any one or combination of memory components RAM 706, ROM 708, or storage device 710, and user input provided via input device 714.

[0103] VI. Illustrative Definitions and Context The present disclosure is not limited to these exemplary embodiments and applications or the manner in which the exemplary embodiments and applications operate or are described herein. Further, the figures may show simplified or partial views, and dimensions of elements in the figures may be exaggerated or out of proportion.

[0104] Additionally, when the terms "on," "attached to," "connected to," "coupled to," or similar terms are used herein, an element (e.g., a component, material, layer, substrate, etc.) may be "on," "attached to," "connected to," or "coupled to" another element, regardless of whether the element is directly on, directly attached to, connected to, or coupled to another element, or whether there are one or more intervening elements between the one element and the other element. Additionally, when a reference is made to a list of elements (e.g., elements a, b, c), such reference is intended to include any one of the listed elements by itself, any combination of less than all of the listed elements, and / or all combinations of the listed elements. The division of sections herein is merely for ease of review and is not intended to limit any combination of the elements described.

[0105] The term "subject" may refer to a subject in a clinical trial, a person receiving treatment, a person undergoing anti-cancer therapy, a person being monitored for remission or recovery, a person undergoing preventive health analysis (e.g., by medical history), or a patient of any other person or subject. In various instances, "subject" and "patient" may be used interchangeably herein.

[0106] Unless otherwise defined, scientific and technical terms used in connection with the present teachings described herein shall have the meanings commonly understood by those skilled in the art.Furthermore, unless otherwise indicated by context, singular terms shall include plurals and plural terms shall include singulars.In general, the nomenclature and techniques utilized in connection with chemistry, biochemistry, molecular biology, pharmacology and toxicology described herein are those well known and commonly used in the art.

[0107] As used herein, "substantially" means sufficient to function for the intended purpose. Thus, the term "substantially" allows for slight, insignificant variations from absolute or perfect conditions, dimensions, measurements, results, etc., as would be expected by one of ordinary skill in the art, but does not noticeably affect overall performance. For parameters or characteristics that are numerical values ​​or can be expressed as numerical values, "substantially" means within 10%.

[0108] The term "a" or "an" means two or more.

[0109] As used herein, the term "plurality" can be 2, 3, 4, 5, 6, 7, 8, 9, 10 or more.

[0110] As used herein, the term "set" means one or more. For example, a set of items includes one or more items.

[0111] As used herein, the phrase "at least one of," when used in conjunction with a list of items, means that one or more different combinations of the listed items may be used, and in some cases, only one of the items in the list is used. The items may be specific objects, things, steps, operations, processes, or categories. In other words, "at least one of" means that any combination or number of items from the list may be used, but not all of the items in the list are used. For example, but not limited to, "at least one of item A, item B, or item C" means item A; item A and item B; item B; item A, item B, item C; item B, item C; or items A and C. In some cases, "at least one of item A, item B, or item C" means, but is not limited to, two of item A, one of item B, and ten of item C; four of item B and seven of item C; or some other suitable combination.

[0112] As used herein, a "model" may include one or more algorithms, one or more mathematical techniques, one or more machine learning algorithms, or a combination thereof.

[0113] As used herein, "machine learning" may be the practice of using algorithms to analyze data, learn from it, and then make decisions or predictions about something in the world. Machine learning uses algorithms that can learn from data without relying on rule-based programming.

[0114] As used herein, "artificial neural network" or "neural network" (NN) may refer to a mathematical algorithm or computational model that mimics an interconnected group of artificial neurons that process information based on a connectionistic approach to computation. A neural network, sometimes referred to as a neural net, may use one or more layers of linear units, nonlinear units, or both to predict an output for a received input. Some neural networks include one or more hidden layers in addition to an output layer. The output of each hidden layer is used as an input to the next layer in the network, i.e., the next hidden layer or the output layer. Each layer of the network generates an output from the received input according to the current value of each parameter set. In various embodiments, a reference to a "neural network" may be a reference to one or more neural networks.

[0115] Neural networks can process information in two ways: when the neural network is being trained, it is in training mode, and when the neural network puts into practice what it has learned, it is in inference (or prediction) mode. Neural networks learn through a feedback process (e.g., backpropagation) that allows the network to adjust (change its behavior) the weight coefficients of individual nodes in the intermediate hidden layers so that the output matches that of the training data. In other words, the neural network learns by being fed training data (training examples), and eventually learns how to arrive at the correct output even when presented with a new range or set of inputs. The neural network may include, for example, but is not limited to, at least one of a feedforward neural network (FNN), a recurrent neural network (RNN), a modular neural network (MNN), a convolutional neural network (CNN), a residual neural network (ResNet), an ordinary differential equation neural network (Neural-ODE), or another type of neural network.

[0116] VII. Enumeration of Exemplary Embodiments Embodiment 1: A method comprising: receiving baseline data of a subject diagnosed with neovascular age-related macular degeneration (nAMD); forming a plurality of predictor inputs for an outcome predictor using the baseline data and regimen data for a plurality of treatment regimens, the plurality of predictor inputs including a different predictor input for each of the plurality of treatment regimens, the outcome predictor including at least one machine learning model; generating a plurality of treatment scores for the plurality of treatment regimens using the plurality of predictor inputs via the outcome predictor; and selecting one of the plurality of treatment regimens as the subject's selected treatment regimen based on the plurality of treatment scores.

[0117] Embodiment 2: The method of embodiment 1, wherein the selected treatment regimen maximizes visual improvement and minimizes injection frequency.

[0118] Embodiment 3: The method of embodiment 1 or 2, wherein the selecting comprises determining whether a single therapeutic regimen of the plurality of therapeutic regimens or multiple therapeutic regimens of the plurality of therapeutic regimens meets criteria of the score.

[0119] Embodiment 4: The method of embodiment 3, wherein the selecting further comprises, in response to a single therapeutic regimen satisfying the criteria of the score, selecting the single therapeutic regimen as the selected therapeutic regimen for the subject, or, in response to a plurality of therapeutic regimens satisfying the criteria of the score, identifying a multiple therapeutic regimen among the plurality of therapeutic regimens that satisfies the set of treatment burden criteria for the subject as the selected therapeutic regimen.

[0120] Embodiment 5: The method of embodiment 4, wherein the set of treatment burden criteria includes at least one of the following: fewest injections, least frequent injections, lowest dosage, lowest drug strength, reduced extent of monitoring, or reduced side effects.

[0121] Embodiment 6: The method of any one of embodiments 1-5, wherein generating comprises independently processing each of the plurality of predictor inputs through an outcome predictor to generate a plurality of treatment scores.

[0122] Embodiment 7: The method of any one of embodiments 1 to 5, wherein the outcome predictor comprises a plurality of predictor models, each of the plurality of predictor models comprising at least one machine learning model, and each of the plurality of predictor models is configured to use a predictor input of the plurality of predictor inputs corresponding to a treatment regimen to generate a treatment score of the plurality of treatment scores for a corresponding treatment regimen of the plurality of treatment regimens.

[0123] Embodiment 8: The method of any one of embodiments 1 to 7, wherein the baseline data includes at least one of clinical data including optical coherence tomography (OCT) image data, visual acuity measurements, central subfield thickness, hypointensity defect, age, or gender, or retinal feature data extracted from segmented image data generated from OCT image data corresponding to the baseline time point.

[0124] Embodiment 9: The method of any one of embodiments 1-8, wherein the regimen data for a corresponding one of the multiple therapeutic regimens specifies a treatment and at least one of an administration frequency, an administration schedule, or a monitoring schedule for the treatment.

[0125] Embodiment 10: The method of any one of embodiments 1 to 9, wherein each of the treatment scores is a predicted visual acuity measurement.

[0126] Embodiment 11: The method of any one of embodiments 1 to 10, wherein the at least one machine learning model comprises at least one of a linear regression model, a random forest (RF) model, a gradient boosting machine (GBM) model, an extreme gradient boosting (XGBoost), or a support vector machine (SVM) model.

[0127] Embodiment 12: A method comprising: receiving clinical data and imaging data for a subject diagnosed with neovascular age-related macular degeneration (nAMD) for a baseline time point; generating retinal characteristic data using the imaging data, the retinal characteristic data comprising at least one of a pathology-related feature, a layer-related volumetric feature, or a layer-related thickness; generating a plurality of treatment scores for a plurality of therapeutic regimens via an outcome predictor comprising at least one machine learning model; and selecting one of the plurality of therapeutic regimens as a selected therapeutic regimen for the subject based on the plurality of treatment scores.

[0128] Embodiment 13: The method of embodiment 12, wherein the selecting comprises determining whether a single therapeutic regimen of the plurality of therapeutic regimens or a multiple therapeutic regimen of the plurality of therapeutic regimens meets criteria of the score.

[0129] Embodiment 14: The method of embodiment 13, wherein the selecting further comprises, in response to a single therapeutic regimen satisfying the criteria of the score, selecting the single therapeutic regimen as the selected therapeutic regimen for the subject, or, in response to a multiple therapeutic regimen satisfying the criteria of the score, identifying a therapeutic regimen among the multiple therapeutic regimens that satisfies a set of treatment burden criteria for the subject as the selected therapeutic regimen.

[0130] Embodiment 15: The method of any one of embodiments 12 to 14, wherein the clinical data includes at least one of visual acuity measurements, central subfield thickness, hypointensity defects, age, or gender at a baseline time point, and the imaging data includes optical coherence tomography (OCT) image data.

[0131] Embodiment 16: The method of any one of embodiments 12 to 15, wherein each of the treatment scores is a predicted visual acuity measurement.

[0132] Embodiment 17: The method of any one of embodiments 12 to 16, wherein the at least one machine learning model comprises at least one of a linear regression model, a random forest (RF) model, a gradient boosting machine (GBM) model, an extreme gradient boosting (XGBoost), or a support vector machine (SVM) model.

[0133] Embodiment 18: A system for predicting a selected treatment regimen for a subject diagnosed with neovascular age-related macular degeneration, comprising: a memory containing a machine-readable medium including machine-executable code; and a processor coupled to the memory, the processor configured to execute machine-executable code that causes the processor to receive baseline data for the subject diagnosed with neovascular age-related macular degeneration (nAMD); form a plurality of predictor inputs for an outcome predictor using the baseline data and regimen data for a plurality of treatment regimens, the plurality of predictor inputs including a different predictor input for each of the plurality of treatment regimens, the outcome predictor including at least one machine learning model; generate a plurality of treatment scores for the plurality of treatment regimens using the plurality of predictor inputs via the outcome predictor; and select one of the plurality of treatment regimens as the selected treatment regimen for the subject based on the plurality of treatment scores.

[0134] Embodiment 19: The system of embodiment 18, wherein the selected treatment regimen maximizes visual improvement and minimizes injection frequency.

[0135] Embodiment 20: The system of embodiment 18 or 19, wherein the processor is further configured to execute machine executable code that causes the processor to determine whether a single therapeutic regimen of the plurality of therapeutic regimens or multiple therapeutic regimens of the plurality of therapeutic regimens meets the criteria of the score, and in response to a single therapeutic regimen meeting the criteria of the score, select the single therapeutic regimen as a selected therapeutic regimen for the subject, or in response to a plurality of therapeutic regimens meeting the criteria of the score, identify a therapeutic regimen of the plurality of therapeutic regimens that meets a set of treatment burden criteria for the subject as the selected therapeutic regimen.

[0136] VIII. Further Considerations Headings and subheadings between sections and subsections in this document are included merely to improve readability and do not imply that features may be combined across the sections and subsections, and thus the sections and subsections do not describe separate embodiments.

[0137] Some embodiments of the present disclosure include a system including one or more data processors. In some embodiments, the system includes a non-transitory computer-readable storage medium including instructions that, when executed on the one or more data processors, cause the one or more data processors to perform some or all of one or more methods and / or some or all of one or more processes disclosed herein. Some embodiments of the present disclosure include a computer program product tangibly embodied in a non-transitory machine-readable storage medium, the computer program product including instructions configured to cause the one or more data processors to perform some or all of one or more methods and / or some or all of one or more processes disclosed herein.

[0138] The terms and expressions used are used as terms of description and not of limitation, and in the use of such terms and expressions there is no intention to exclude equivalents of the features shown and described or portions thereof, but it is recognized that various modifications are possible within the scope of the invention as claimed. Thus, although the invention as claimed has been specifically disclosed by embodiments and optional features, it is to be understood that modifications and variations of the concepts disclosed herein may be employed by those skilled in the art, and such modifications and variations are considered to be within the scope of the invention as defined by the appended claims.

[0139] The following description provides only preferred exemplary embodiments and is not intended to limit the scope, applicability or configuration of the present disclosure. Rather, the following description of preferred exemplary embodiments provides those skilled in the art with an enabling description for implementing various embodiments. It will be understood that various changes may be made in the function and arrangement of elements (e.g., block or schematic elements, flow diagram elements, etc.) without departing from the spirit and scope of the appended claims.

[0140] Specific details are given in the following description to provide a thorough understanding of the embodiments. However, it will be understood that the embodiments may be practiced without these specific details. For example, circuits, systems, networks, processes, and other components may be shown as components in block diagram form in order to avoid obscuring the embodiments in unnecessary detail. In other instances, well-known circuits, processes, algorithms, structures, and techniques may be shown without unnecessary detail in order to avoid obscuring the embodiments.

[0141] While the present teachings have been described in conjunction with various embodiments, it is not intended that the present teachings be limited to such embodiments. On the contrary, the present teachings encompass various alternatives, modifications, and equivalents, as will be appreciated by those skilled in the art.

[0142] In describing various embodiments, the specification may present a method and / or process as a particular sequence of steps. However, to the extent that the method or process does not rely on the particular order of steps described herein, the method or process should not be limited to the particular order of steps described, and one of ordinary skill in the art can readily appreciate that the order may be altered and still be within the spirit and scope of the various embodiments.

Claims

1. Receiving baseline data for a subject diagnosed with neovascular age-related macular degeneration (nAMD); forming a plurality of predictor inputs for an outcome predictor using said baseline data and regimen data for a plurality of treatment regimens; the plurality of predictor inputs comprises a different predictor input for each of the plurality of treatment regimens; forming a plurality of predictor inputs, the outcome predictors including at least one machine learning model; generating, via the outcome predictor, a plurality of treatment scores for the plurality of therapeutic regimens using the plurality of predictor inputs; and selecting one of the plurality of therapeutic regimens as a selected therapeutic regimen for the subject based on the plurality of treatment scores; A method comprising:

2. 10. The method of claim 1, wherein the selected treatment regimen maximizes visual improvement and minimizes injection frequency.

3. 3. The method of claim 1 or 2, wherein said selecting comprises determining whether a single therapeutic regimen of said plurality of therapeutic regimens or multiple therapeutic regimens of said plurality of therapeutic regimens meets a criterion of a score.

4. The selecting, in response to the single therapeutic regimen meeting the criteria of the score, selecting the single therapeutic regimen as the selected therapeutic regimen for the subject; or 4. The method of claim 3, comprising, in response to the multi-treatment regimen satisfying criteria of the score, identifying a treatment regimen among the multi-treatment regimen that satisfies a set of treatment burden criteria for the subject as the selected treatment regimen.

5. 5. The method of claim 4, wherein the set of treatment burden criteria comprises at least one of: fewest number of injections, least frequent injections, lowest dosage, lowest drug strength, reduced degree of monitoring, or reduced side effects.

6. 6. The method of any one of claims 1-5, wherein said generating comprises independently processing each of the plurality of predictor inputs through the outcome predictor to generate the plurality of treatment scores.

7. The outcome predictor comprises a multiple predictor model; each of the plurality of predictor models comprises at least one machine learning model; 6. The method of any one of claims 1 to 5, wherein each of the plurality of predictor models is configured to generate a treatment score of the plurality of treatment scores for a corresponding treatment regimen of the plurality of treatment regimens using a predictor input of the plurality of predictor inputs corresponding to the treatment regimen.

8. The baseline data includes: Optical coherence tomography (OCT) image data; Clinical data including at least one of visual acuity measurements, central subfield thickness, hypoluminal defect, age, or sex; or The method of any one of claims 1 to 7, comprising at least one of retinal feature data extracted from segmented image data generated from OCT image data corresponding to a baseline time point.

9. 9. The method of any one of claims 1-8, wherein the regimen data for a corresponding one of the plurality of therapeutic regimens specifies a treatment and at least one of an administration frequency, an administration schedule, or a monitoring schedule for the treatment.

10. The method of any one of claims 1 to 9, wherein each of the treatment scores is a predicted visual acuity measurement.

11. 11. The method of any one of claims 1 to 10, wherein the at least one machine learning model comprises at least one of a linear regression model, a random forest (RF) model, a gradient boosting machine (GBM) model, an extreme gradient boosting (XGBoost), or a support vector machine (SVM) model.

12. receiving clinical and imaging data for a subject diagnosed with neovascular age-related macular degeneration (nAMD) for a baseline time point; generating retinal feature data using the imaging data, the retinal feature data including at least one of a pathology-related feature, a layer-related volumetric feature, or a layer-related thickness; generating a plurality of treatment scores for a plurality of therapeutic regimens via an outcome predictor comprising at least one machine learning model; selecting one of the plurality of therapeutic regimens as a selected therapeutic regimen for the subject based on the plurality of treatment scores; A method comprising:

13. 13. The method of claim 12, wherein said selecting comprises determining whether a single therapeutic regimen of said plurality of therapeutic regimens or multiple therapeutic regimens of said plurality of therapeutic regimens meets a criterion of a score.

14. The selecting, in response to the single therapeutic regimen meeting the criteria of the score, selecting the single therapeutic regimen as the selected therapeutic regimen for the subject; or 14. The method of claim 13, further comprising, in response to the multi-treatment regimen meeting criteria of the score, identifying a treatment regimen among the multi-treatment regimen that meets a set of treatment burden criteria for the subject as the selected treatment regimen.

15. 15. The method of any one of claims 12-14, wherein the clinical data comprises at least one of a visual acuity measurement, central subfield thickness, hypointensity defect, age, or gender for the baseline time point, and the imaging data comprises optical coherence tomography (OCT) image data.

16. The method of any one of claims 12 to 15, wherein each of the treatment scores is a predicted visual acuity measurement.

17. 17. The method of any one of claims 12 to 16, wherein the at least one machine learning model comprises at least one of a linear regression model, a random forest (RF) model, a gradient boosting machine (GBM) model, an extreme gradient boosting (XGBoost), or a support vector machine (SVM) model.

18. 1. A system for predicting a selected treatment regimen for a subject diagnosed with neovascular age-related macular degeneration, comprising: a memory containing a machine-readable medium including machine-executable code; and a processor coupled to the memory; The processor may include: Receiving baseline data for a subject diagnosed with neovascular age-related macular degeneration (nAMD); forming a plurality of predictor inputs for an outcome predictor using said baseline data and regimen data for a plurality of treatment regimens; the plurality of predictor inputs comprises a different predictor input for each of the plurality of treatment regimens; forming a plurality of predictor inputs, the outcome predictors including at least one machine learning model; generating, via the outcome predictor, a plurality of treatment scores for the plurality of therapeutic regimens using the plurality of predictor inputs; and selecting one of the plurality of therapeutic regimens as a selected therapeutic regimen for the subject based on the plurality of treatment scores; A system configured to execute the machine-executable code to cause a

19. 20. The system of claim 18, wherein the selected treatment regimen maximizes visual improvement and minimizes injection frequency.

20. The processor may include: determining whether a single treatment regimen of the plurality of treatment regimens or multiple treatment regimens of the plurality of treatment regimens meets a criterion of a score; and in response to the single therapeutic regimen meeting the criteria of the score, selecting the single therapeutic regimen as the selected therapeutic regimen for the subject; or 20. The system of claim 18 or 19, further configured to execute the machine executable code to: in response to the multi-treatment regimen satisfying criteria of the score, identify a treatment regimen among the multi-treatment regimen that satisfies a set of treatment burden criteria for the subject as the selected treatment regimen.

Images