System and method for processing retinal signal data and identifying conditions

Abstract

A method and system for predicting the likelihood that a patient has one or more conditions is disclosed. Retinal signal data corresponding to the patient may be received. Retinal signal features may be extracted from the retinal signal data. Descriptors may be extracted from the retinal signal features. The descriptors may be applied to a first mathematical model and a second mathematical model. The first mathematical model may correspond to a first condition. The second mathematical model may correspond to a second condition. A first predicted probability for the first condition may be generated. A second predicted probability for the second condition may be generated. The first predicted probability and the second predicted probability may be output.

Description

Technical Field

[0001] Cross - reference to Related Applications This application claims the benefit of U.S. Provisional Patent Application No. 63 / 000,055, filed Mar. 26, 2020; U.S. Provisional Patent Application No. 63 / 038,257, filed Jun. 12, 2020; and U.S. Provisional Patent Application No. 63 / 149,508, filed Feb. 15, 2021, each of which is incorporated herein by reference in its entirety.

[0002] This technology relates to systems and methods for processing retinal signal data generated by light stimulation.

Background Art

[0003] Clinicians may want to determine whether a patient is in a medical state, such as a psychiatric or neurological state. A clinician can compare a patient to known criteria to determine what state the patient is in. In some cases, a patient may fit multiple states, and it may be difficult or impossible for a clinician to distinguish between the states. It may be desirable for a clinician to have tools that assist in determining and / or confirming whether a patient is in a medical state and / or in distinguishing between those states.

Summary of the Invention

Problems to be Solved by the Invention

[0004] An object of this technology is to improve at least some of the limitations that exist in the prior art.

Means for Solving the Problems

[0005] Embodiments of this technology were developed based on the developers' recognition of certain drawbacks associated with existing systems for determining medical states.

[0006] For example, identifying biomarkers and / or biosignatures for psychiatric, mental, or neurological conditions such as schizophrenia, bipolar I disorder, or depression can enable healthcare professionals to make earlier diagnoses of a condition, identify which condition a patient has when multiple candidate conditions exist, and / or initiate early and potentially preventive interventions. This early diagnose of a medical condition can improve patient treatment and / or prognosis.

[0007] Embodiments of this technology were developed based on the developers' observation that data acquired in electroretinography (ERG) can provide some insight into determining a medical condition. However, existing methods for collecting and analyzing ERGs can only collect and analyze a limited amount of information from the captured electrical signals. It has been found that expanding the amount of information collected regarding the retina's response to light stimuli has made it possible to generate retinal signal data with higher density information, a greater volume of information, and / or additional types of information. This retinal signal data enables multimodal mapping of electrical signals and / or other data, allowing for the detection of additional features in specific multimodal mappings for specific conditions. Multimodal mapping may include multiple parameters of the retinal signal data, such as time, frequency, light stimulus parameters, and / or any other parameters.

[0008] Embodiments of this technology possess a greater volume of information, higher density of information, and / or additional types of information detail compared to conventional ERG data, forming the basis for a sophisticated methodology for determining medical conditions based on the processing of retinal signal data, which is referred to herein as “Retinal Signal Processing and Analysis” (RSPA). In certain embodiments, this retinal signal data enables mathematical modeling of datasets containing various types of information, identification of retinal signal features, and the ability to identify biomarkers and / or biosignatures within the retinal signal data using, for example, retinal signal features. Certain non-essential embodiments of this technology also provide methods for collecting retinal signal data that possess a greater volume of information, higher density of information, and / or additional types of information compared to ERG data.

[0009] In certain embodiments of this technology, more accurate detection of specific medical conditions, or more distinguishable separation between medical conditions, may be achieved. Detection of medical conditions or distinguishing between conditions may be achieved within a diversity of relevant factors (e.g., sex, age, disease onset, retinal pigmentation, iris color, etc.) and / or confounding factors (onset of condition, drug use, effects of certain treatments, episodes of psychosis, anxiety, depression, overlapping signs and symptoms common to several disorders). An increase in the level of detail and / or number of retinal signal features that the technology can capture and analyze has a direct impact on the possibility of using retinal signal data to identify biosignatures, to better distinguish between conditions such as pathological conditions, and to better describe conditions, compared to controls (sometimes referred to as non-pathological conditions). For example, embodiments of this method may be based on retinal signal data captured at higher sampling frequencies and / or longer periods compared to conventional ERGs. Retinal signal data may include additional features recorded along with electrical signals, such as impedance, light wavelength, light spectrum, or light intensity reaching the retina, but are not limited to these. The capture of additional information at higher sampling frequencies, and / or data collected with extended-range retinal light stimulation over longer periods, as well as its multidimensional representation, may be referred to as “high-density retinal signal data.” High-density captured retinal signal data can contain more information than data previously captured during ERG (referred to as “conventional ERG”). Unlike conventional ERG, retinal signal data can be voltage-independent and / or time-independent.

[0010] In certain embodiments, high-density retinal signal data can be used to enable more efficient processing of retinal signal data. The advantage of high-density retinal signal data compared to conventional ERG data is that it benefits from a large amount of information about electrical signals and additional retinal signal features, and therefore from more detailed biosignatures. Thus, using a higher level of detail makes it possible to better distinguish between different states and prepare a set of classifiers that represent the biosignature features of each state.

[0011] According to a first broad aspect of the present technology, a method is provided for generating a mathematical model corresponding to a first state, the method being executable by at least one processor of a computer system, the method comprising the step of generating a mathematical model corresponding to a first state, the method being executable by at least one processor of a computer system, the method comprising the step of receiving a plurality of datasets of labeled retinal signal data corresponding to a plurality of patients, each dataset comprising retinal signal data and labels of a patient, the labels indicating whether the patient is in a first state, the step of extracting a set of features from the retinal signal data, the step of selecting a subset of features from the set of features, the subset of features corresponding to a biomarker of the first state, and the step of determining one or more classifiers that distinguish the first state from a second state based on the subset of features.

[0012] In some implementations of the method, the set of features includes voltage, circuit impedance, signal acquisition time, sampling frequency, photostimulus synchronization time, photostimulus offset, or an indicator of which retinal region was illuminated.

[0013] In some implementations of the method, the set of features includes eye position, pupil size, applied brightness intensity, frequency of light stimulation, frequency of retinal signal sampling, illumination wavelength, illumination time, background wavelength, or background brightness.

[0014] In some implementations of the method, one or more classifiers distinguish the biosignature of a first state from the biosignature of a second state.

[0015] In some implementations of the method, the method further includes a step of ranking a set of features based on the relevance of each feature to the biosignature of a first state, and a step of selecting a subset of features includes a step of selecting the highest-ranked feature from the set of features.

[0016] In some implementations of the method, the method further includes the steps of receiving clinical information adjuvants corresponding to multiple patients, wherein each dataset contains clinical information adjuvants for the patients, and selecting a subset of clinical adjuvants, wherein the clinical adjuvants within the subset influence the detection of biomarkers.

[0017] In some implementations of the method, clinical information cofactors include each patient's age, sex, skin pigmentation, or iris color.

[0018] In some implementations of the method, the step of determining one or more classifiers includes the step of determining one or more classifiers based on a subset of clinical cofactors.

[0019] In some implementations of the method, the method further includes the step of generating a mathematical model based on one or more classifiers.

[0020] In some implementations of the method, the method further includes the steps of inputting patient-dependent retinal signal data and clinical information cofactors into a mathematical model, and having the mathematical model output the predictive likelihood that the patient is in a first state.

[0021] In some implementations of the method, the method further includes the steps of inputting patient-dependent retinal signal data and clinical information cofactors into a mathematical model, and having the mathematical model output the predictive likelihood that the patient is not in a first state.

[0022] In some implementations of the method, the first condition is schizophrenia, bipolar disorder, major depressive disorder, or psychosis.

[0023] In some implementations of the method, the first state is post-traumatic stress disorder, stroke, drug abuse, obsessive-compulsive disorder, Alzheimer's disease, Parkinson's disease, multiple sclerosis, autism, or attention deficit disorder.

[0024] In some implementations of the method, the retinal signal data has a sampling frequency between 4 and 24 kHz.

[0025] In some implementations of the method, the retinal signal data is collected during a signal acquisition time between 200 milliseconds and 500 milliseconds.

[0026] In some implementations of the method, the retinal signal data includes the impedance component of the receiving circuit continuously recorded while capturing the retinal signal data.

[0027] In some implementations of the method, the retinal signal data includes one or more optical parameters.

[0028] In some implementations of the method, the optical parameters include the luminance of the retinal light stimulation or the pupil size.

[0029] According to another broad aspect of the present technique, a method for predicting the probability that a patient is in one or more states is provided, the method being executable by at least one processor of a computer system, the method comprising receiving retinal signal data corresponding to the patient, extracting one or more retinal signal features from the retinal signal data, extracting one or more descriptors from the retinal signal features, and applying the one or more descriptors to a first mathematical model and a second mathematical model, wherein the first mathematical model corresponds to a first state and the second mathematical model corresponds to a second state, thereby generating a first prediction probability for the first state and a second prediction probability for the second state, and outputting the first prediction probability and the second prediction probability.

[0030] In some implementations of the method, the method further includes the step of displaying an interface that includes a first predicted probability and a second predicted probability.

[0031] In some implementations of the method, the method further includes the step of storing a first predicted probability and a second predicted probability.

[0032] In some implementations of the method, the method further includes the step of collecting retinal signal data.

[0033] In some implementations of the method, the method further includes the steps of obtaining clinical information cofactors extracted from clinical information corresponding to a patient, and applying the clinical information cofactors to a first mathematical model and a second mathematical model.

[0034] In some implementations of the method, clinical information cofactors correspond to the patient's age, sex, skin pigmentation, or iris color.

[0035] In some implementations of the method, the retinal signal data has a sampling frequency between 4 and 24 kHz.

[0036] In several implementations of the method, retinal signal data is collected during a signal acquisition time of 200 to 500 milliseconds.

[0037] In some implementations of the method, the retinal signal data includes the impedance component of the receiving circuit, which is continuously recorded while the retinal signal data is being captured.

[0038] In some implementations of the method, the retinal signal data includes one or more optical parameters.

[0039] In some implementations of the method, the optical parameters include the luminance of the retinal light stimulus or the pupil size.

[0040] In some implementations of the method, the first state is a medical state, and the second state is a control state.

[0041] In some implementations of the method, the first or second state is schizophrenia, bipolar disorder, major depressive disorder, or psychosis.

[0042] In some implementations of the method, the first or second state is post-traumatic stress disorder, stroke, substance abuse, obsessive-compulsive disorder, Alzheimer's disease, Parkinson's disease, multiple sclerosis, autism, or attention deficit disorder.

[0043] In some implementations of the method, the method further includes the step of receiving user input indicating a selection of a first state and a second state.

[0044] In some implementations of the method, the method further includes the steps of selecting a drug based on a first and a second predicted probability, and administering the drug to a patient.

[0045] According to another broader aspect of the present technique, a method is provided for determining a biosignature of a condition, the method being executable by at least one processor of a computer system, the method comprising: receiving a plurality of datasets of labeled retinal signal data corresponding to a plurality of patients, each dataset comprising retinal signal data and labels of a patient, the labels indicating whether the patient is in a condition; extracting a set of features from the retinal signal data; selecting a subset of features from the set of features, the subset of features corresponding to a biomarker of the condition; and determining one or more classifiers that identify a biosignature of the condition based on the subset of features.

[0046] In some implementations of the method, the method further includes the steps of receiving clinical information cofactors corresponding to multiple patients, wherein each dataset contains the clinical information cofactors of the patients; selecting a subset of clinical cofactors, wherein the clinical cofactors within the subset influence the detection of biomarkers; and determining one or more classifiers, which includes determining one or more classifiers based on the subset of clinical cofactors.

[0047] In some implementations of the method, the method further includes a step of ranking a set of features based on the relevance of each feature to the biosignature of a first state, and a step of selecting a subset of features includes a step of selecting the highest-ranked feature from the set of features.

[0048] According to another broad aspect of the present technology, a system is provided for predicting the probability that a patient is in one or more states, the system comprising an optical stimulator, one or more sensors, and a computer system, the computer system comprising at least one processor and memory, the memory storing a plurality of executable instructions, when executed by at least one processor, causing the system to cause the optical stimulator to provide an optical stimulator signal to the patient's retina; to collect electrical signals in response to the optical stimulator via one or more sensors; to generate retinal signal data corresponding to the patient based on the electrical signals; to extract one or more retinal signal features from the retinal signal data; to extract one or more descriptors from the retinal signal features; and to apply one or more descriptors to a first mathematical model and a second mathematical model, the first mathematical model corresponding to a first state and the second mathematical model corresponding to a second state, thereby generating a first predicted probability for the first state and a second predicted probability for the second state, and to output the first predicted probability and the second predicted probability.

[0049] In some implementations of the system, retinal signal data includes optical wavelength components recorded while the retinal signal data is being captured.

[0050] In some implementations of the system, retinal signal data includes optical spectral components recorded while capturing retinal signal data.

[0051] In some implementations of the system, retinal signal data includes light intensity components recorded while capturing retinal signal data.

[0052] In some implementations of the system, retinal signal data includes an illuminated retinal surface component recorded while the retinal signal data is being captured. The retinal surface component may represent the surface area of ​​the illuminated retina.

[0053] According to another broad aspect of the present technology, a system is provided for predicting the probability that a patient is in one or more states, the system comprising a computer system comprising at least one processor and memory, the memory storing a plurality of executable instructions that, when executed by at least one processor, cause the computer system to perform the following actions: receive retinal signal data corresponding to a patient; extract one or more retinal signal features from the retinal signal data; extract one or more descriptors from the retinal signal features; apply one or more descriptors to a first mathematical model and a second mathematical model, the first mathematical model corresponding to a first state and the second mathematical model corresponding to a second state, thereby generating a first predicted probability for the first state and a second predicted probability for the second state, and outputting the first predicted probability and the second predicted probability.

[0054] In some implementations of the system, the system further comprises a photostimulator and one or more sensors, and when an instruction is executed by at least one processor, the computer system is instructed to cause the photostimulator to provide a photostimulator signal to the patient's retina, to acquire an electrical signal in response to the photostimulator via one or more sensors, and to generate retinal signal data based on the electrical signal.

[0055] In some implementations of the system, the system further includes a display, and when an instruction is executed by at least one processor, the system causes the system to output an interface via the display that includes a first predicted probability and a second predicted probability.

[0056] According to another broad aspect of the present technology, a method is provided for monitoring the condition of a patient, the method being executable by at least one processor of a computer system, the method comprising: receiving retinal signal data corresponding to a patient; extracting one or more retinal signal features from the retinal signal data; extracting one or more descriptors from the retinal signal features; applying one or more descriptors to a mathematical model corresponding to a state, thereby generating a predictive probability about the state; and outputting the predictive probability.

[0057] In some implementations of the method, the method further includes the steps of selecting a drug based on a predicted probability and administering the drug to a patient.

[0058] In some implementations of the method, retinal signal data includes retinal signal data captured during the treatment of a patient for a condition.

[0059] In some implementations of the method, the condition is schizophrenia, bipolar disorder, major depressive disorder, psychosis, post-traumatic stress disorder, stroke, substance abuse, obsessive-compulsive disorder, Alzheimer's disease, Parkinson's disease, multiple sclerosis, autism, or attention deficit disorder.

[0060] In the context of this specification, unless expressly otherwise specified, computer systems may refer to, but are not limited to, “electronic devices,” “operating systems,” “systems,” “computer-based systems,” “controller units,” “control devices,” and / or any combination thereof appropriate for the relevant task at hand.

[0061] In the context of this specification, unless expressly otherwise specified, the terms “computer-readable media” and “memory” are intended to include media of all natures and types, non-exclusive examples of which include RAM, ROM, disks (such as CD-ROMs, DVDs, floppy disks, and hard disk drives), USB keys, flash memory cards, solid-state drives, and tape drives.

[0062] In the context of this specification, “database” is any structured collection of data, regardless of its particular structure, database management software, or computer hardware on which the data is stored, implemented, or otherwise made available. A database may reside on the same hardware as the processes that store or utilize the information stored within it, or it may reside on separate hardware, such as a dedicated server or multiple servers.

[0063] In the context of this specification, unless otherwise explicitly stated, words such as “first,” “second,” and “third” are used as adjectives solely for the purpose of distinguishing the nouns they modify from one another, and not for the purpose of describing a specific relationship between those nouns.

[0064] Each embodiment of the present technology has at least one of the purposes and / or aspects described above, but not necessarily all of them. It should be understood that some embodiments of the present technology resulting from an attempt to achieve the purposes described above may not satisfy these purposes and / or may satisfy other purposes not specifically described herein.

[0065] Additional and / or alternative features, aspects, and advantages of embodiments of this technology will become apparent from the following description, accompanying drawings, and accompanying claims.

[0066] For a better understanding of this technology, as well as other embodiments and further features, refer to the following description used in conjunction with the accompanying drawings. [Brief explanation of the drawing]

[0067] [Figure 1] This is a block diagram illustrating an exemplary computing environment according to various embodiments of this technology. [Figure 2] This is a block diagram of a retinal signal processing system according to various embodiments of this technology. [Figure 3] This figure shows an exemplary functional architecture of an information processing method that leads to the construction of a mathematical function for predicting whether a patient is in a certain condition, according to various embodiments of this technology. [Figure 4] This is a flowchart illustrating a method for predicting the likelihood of a medical condition using various embodiments of this technology. [Figure 5] This is a flowchart illustrating a method for generating mathematical models to predict whether a patient is in a certain condition, based on various embodiments of this technology. [Figure 6] This is a flowchart illustrating a method for training a machine learning algorithm (MLA) to predict the likelihood of a medical condition, using various embodiments of this technology. [Figure 7]This is a flowchart illustrating methods for using MLA to predict the likelihood of a medical condition, according to various embodiments of this technology. [Figure 8] This figure shows examples of time-frequency analysis and selection of a distinguishing region based on the statistical significance of higher-magnitude frequencies, according to various embodiments of this technology. [Figure 9] This figure shows examples of time-frequency analysis and selection of a distinguishing region based on the statistical significance of higher-magnitude frequencies, according to various embodiments of this technology. [Figure 10] This figure shows examples of time-frequency analysis and selection of a distinguishing region based on the statistical significance of higher-magnitude frequencies, according to various embodiments of this technology. [Figure 11] This figure shows examples of time-frequency analysis and selection of a distinguishing region based on the statistical significance of higher-magnitude frequencies, according to various embodiments of this technology. [Figure 12] This figure shows examples of time-frequency analysis and selection of a distinguishing region based on the statistical significance of higher-magnitude frequencies, according to various embodiments of this technology. [Figure 13] This figure shows examples of time-frequency analysis and selection of a distinguishing region based on the statistical significance of higher-magnitude frequencies, according to various embodiments of this technology. [Figure 14] This figure shows examples of time-frequency analysis and selection of a distinguishing region based on the statistical significance of higher-magnitude frequencies, according to various embodiments of this technology. [Figure 15] This figure shows examples of time-frequency analysis and selection of a distinguishing region based on the statistical significance of higher-magnitude frequencies, according to various embodiments of this technology. [Figure 16] This figure shows examples of time-frequency analysis and selection of a distinguishing region based on the statistical significance of higher-magnitude frequencies, according to various embodiments of this technology. [Figure 17] This figure shows examples of time-frequency analysis and selection of a distinguishing region based on the statistical significance of higher-magnitude frequencies, according to various embodiments of this technology. [Figure 18]This figure shows examples of time-frequency analysis and selection of a distinguishing region based on the statistical significance of higher-magnitude frequencies, according to various embodiments of this technology. [Figure 19] This figure shows examples of time-frequency analysis and selection of a distinguishing region based on the statistical significance of higher-magnitude frequencies, according to various embodiments of this technology. [Figure 20] This figure shows examples of time-frequency analysis and selection of a distinguishing region based on the statistical significance of higher-magnitude frequencies, according to various embodiments of this technology. [Figure 21] This figure shows examples of time-frequency analysis and selection of a distinguishing region based on the statistical significance of higher-magnitude frequencies, according to various embodiments of this technology. [Figure 22] This figure shows examples of time-frequency analysis and selection of a distinguishing region based on the statistical significance of higher-magnitude frequencies, according to various embodiments of this technology. [Figure 23] This figure shows examples of time-frequency analysis and selection of a distinguishing region based on the statistical significance of higher-magnitude frequencies, according to various embodiments of this technology. [Figure 24] This figure shows examples of time-frequency analysis and selection of a distinguishing region based on the statistical significance of higher-magnitude frequencies, according to various embodiments of this technology. [Figure 25] This figure shows examples of time-frequency analysis and selection of a distinguishing region based on the statistical significance of higher-magnitude frequencies, according to various embodiments of this technology. [Figure 26] This figure shows examples of time-frequency analysis and selection of a distinguishing region based on the statistical significance of higher-magnitude frequencies, according to various embodiments of this technology. [Figure 27] This figure shows examples of time-frequency analysis and selection of a distinguishing region based on the statistical significance of higher-magnitude frequencies, according to various embodiments of this technology. [Figure 28] This figure shows examples of retinal signal feature selection and statistical significance mapping of selected retinal signal features according to various embodiments of this technology. [Figure 29]This figure shows examples of retinal signal feature selection and statistical significance mapping of selected retinal signal features according to various embodiments of this technology. [Figure 30] This figure shows examples of retinal signal feature selection and statistical significance mapping of selected retinal signal features according to various embodiments of this technology. [Figure 31] This figure shows examples of retinal signal feature selection and statistical significance mapping of selected retinal signal features according to various embodiments of this technology. [Figure 32] This figure shows examples of retinal signal feature selection and statistical significance mapping of selected retinal signal features according to various embodiments of this technology. [Figure 33] This figure shows examples of retinal signal feature selection and statistical significance mapping of selected retinal signal features according to various embodiments of this technology. [Figure 34] This figure shows examples of retinal signal feature selection and statistical significance mapping of selected retinal signal features according to various embodiments of this technology. [Figure 35] This figure shows examples of retinal signal feature selection and statistical significance mapping of selected retinal signal features according to various embodiments of this technology. [Figure 36] This figure shows examples of retinal signal feature selection and statistical significance mapping of selected retinal signal features according to various embodiments of this technology. [Figure 37] This figure shows examples of retinal signal feature selection and statistical significance mapping of selected retinal signal features according to various embodiments of this technology. [Figure 38] This figure shows examples of retinal signal feature selection and statistical significance mapping of selected retinal signal features according to various embodiments of this technology. [Figure 39] This figure shows examples of retinal signal feature selection and statistical significance mapping of selected retinal signal features according to various embodiments of this technology. [Figure 40]This figure shows examples of retinal signal feature selection and statistical significance mapping of selected retinal signal features according to various embodiments of this technology. [Figure 41] This figure shows examples of retinal signal feature selection and statistical significance mapping of selected retinal signal features according to various embodiments of this technology. [Figure 42] This figure shows examples of retinal signal feature selection and statistical significance mapping of selected retinal signal features according to various embodiments of this technology. [Figure 43] This figure shows examples of descriptor selection and mapping based on the magnitude of the discriminative power between two states, according to various embodiments of this technology. [Figure 44] This figure shows examples of descriptor selection and mapping based on the magnitude of the discriminative power between two states, according to various embodiments of this technology. [Figure 45] This figure shows examples of descriptor selection and mapping based on the magnitude of the discriminative power between two states, according to various embodiments of this technology. [Figure 46] This figure shows examples of descriptor selection and mapping based on the magnitude of the discriminative power between two states, according to various embodiments of this technology. [Figure 47] This figure shows examples of descriptor selection and mapping based on the magnitude of the discriminative power between two states, according to various embodiments of this technology. [Figure 48] This figure shows examples of descriptor selection and mapping based on the magnitude of the discriminative power between two states, according to various embodiments of this technology. [Figure 49] This figure shows examples of descriptor selection and mapping based on the magnitude of the discriminative power between two states, according to various embodiments of this technology. [Figure 50] This figure shows examples of descriptor selection and mapping based on the magnitude of the discriminative power between two states, according to various embodiments of this technology. [Figure 51] This figure shows examples of descriptor selection and mapping based on the magnitude of the discriminative power between two states, according to various embodiments of this technology. [Figure 52] This figure shows examples of descriptor selection and mapping based on the magnitude of the discriminative power between two states, according to various embodiments of this technology. [Figure 53] This figure shows examples of descriptor selection and mapping based on the magnitude of the discriminative power between two states, according to various embodiments of this technology. [Figure 54] This figure shows examples of descriptor selection and mapping based on the magnitude of the discriminative power between two states, according to various embodiments of this technology. [Figure 55] This figure shows examples of descriptor selection and mapping based on the magnitude of the discriminative power between two states, according to various embodiments of this technology. [Figure 56] This figure shows examples of descriptor selection and mapping based on the magnitude of the discriminative power between two states, according to various embodiments of this technology. [Figure 57] This figure shows examples of descriptor selection and mapping based on the magnitude of the discriminative power between two states, according to various embodiments of this technology. [Figure 58] This figure shows examples of descriptor selection and mapping based on the magnitude of the discriminative power between two states, according to various embodiments of this technology. [Modes for carrying out the invention]

[0068] Unless explicitly stated otherwise, it should be noted that drawings are not to a fixed scale.

[0069] Certain aspects and embodiments of this technology are directed toward methods and systems for processing retinal signals. Generally, certain aspects and embodiments of this technology are based on higher-density retinal signal data, which can be acquired in any way, for example, by increasing the conditions relating to the light stimulus (e.g., the number and range of light intensities), adding the dynamic resistance (impedance) of the circuit used to collect the electrical signals, capturing retinal signal data for longer periods of time, and / or capturing retinal signal data at a higher frequency. To minimize, reduce, or circumvent the limitations pointed out in the prior art, the present computer implementation method is provided for analyzing retinal signal data and extracting retinal signal features that can be used in combination to further decode biomarkers and / or biosignatures. In certain optional embodiments, methods and systems for capturing high-density retinal signal data are provided.

[0070] Certain aspects and embodiments of this technology provide methods and systems that can analyze retinal signal data and provide predictive likelihoods for specific conditions, taking into account several different clinical information cofactors. The conditions may be psychiatric, psychological, neurological, and / or any other type of medical condition. The predictive likelihood may indicate that a patient is currently in a condition and / or is at risk of developing the condition. For example, if a patient's parent is in a condition, the patient's retinal signal data may be analyzed to determine whether the patient is likely to be in the same condition as their parent.

[0071] The systems and methods described herein may be fully or at least partially automated to minimize clinician input when determining a medical condition or a treatment plan for a medical condition. The predictions output by the systems and methods described herein may be used by clinicians as an aid when determining a medical condition and / or when developing a treatment plan for a patient.

[0072] The systems and methods described herein may include: 1) collecting retinal signal data from a patient; 2) labeling each patient's retinal signal data with a label indicating the patient's potential medical condition (which may have been diagnosed by a clinician); 3) extracting retinal signal features from the retinal signal data; 4) selecting a subset of features corresponding to biomarkers of the condition; and / or 5) generating a mathematical model for identifying the patient's distance from biosignatures of a condition by determining a classifier that distinguishes the biosignatures of a condition from biosignatures of other conditions. Biosignatures may include portions of retinal signal data specific to a condition. Retinal signal data may include several biosignatures, each biosignature being specific to a condition. Conditions may include, but are not limited to, schizophrenia, bipolar disorder, major depressive disorder, psychosis, post-traumatic stress disorder, stroke, substance abuse, obsessive-compulsive disorder, Alzheimer's disease, Parkinson's disease, multiple sclerosis, autism, attention deficit disorder, and / or any other condition. These steps can be used to construct mathematical models of any state having biomarkers embedded within retinal signal data. The steps for collecting retinal signal data described herein may be applicable to the analysis of features specific to any state represented in the retinal signal data.

[0073] The systems and methods described herein may include: 1) collecting patient retinal signal data; 2) receiving a selection of conditions to be examined; 3) extracting a mathematical model corresponding to the selected conditions; 4) extracting retinal signal features from the retinal signal data; 5) extracting descriptors from the retinal signal features such that the descriptors relate to the biosignatures of the selected conditions in the retinal signal data; 6) applying the descriptors to the mathematical model; and / or 7) predicting whether a patient is in a condition by outputting the predicted probability that the patient is in each condition. Clinical information of the patient may be collected. Clinical information cofactors may be generated using the clinical information. Clinical information cofactors may also be applied to the mathematical model.

[0074] The systems and methods described herein are based on retinal signal data that has a higher level of information compared to data captured by conventional ERGs. The collected retinal signal data can be analyzed using mathematical and statistical calculations to extract specific retinal signal features. Retinal signal features may include parameters of the retinal signal data and / or features generated using the retinal signal data. Descriptors can be extracted from the retinal signal features. A graphical representation of the findings can be created and output, providing visual support for the selections made when selecting the relevant retinal signal features and / or descriptors. The application can apply mathematical and / or statistical analysis of the results, enabling quantification of those retinal signal features and / or descriptors and comparisons between various states. Based on the retinal signal data and / or any other clinical information, classifiers can be constructed that describe the biosignatures of the states identified in the retinal signal data. Patient retinal signal data can be collected, and the distance between the patient's retinal signal data and the identified biosignatures can be determined, for example, by using the classifiers.

[0075] Computing environment Figure 1 shows a computing environment 100 that may be used to implement and / or run any of the methods described herein. In some embodiments, the computing environment 100 may be implemented by any of the following: a conventional personal computer, a network device and / or an electronic device (but not limited to mobile devices, tablet devices, servers, controller units, control devices, etc.), and / or any combination thereof suitable for the relevant task at hand. In some embodiments, the computing environment 100 comprises various hardware components including one or more single or multicore processors collectively represented by a processor 110, a solid-state drive 120, random-access memory 130, and an input / output interface 150. The computing environment 100 may be a computer specifically designed to run machine learning algorithms (MLAs). The computing environment 100 may be a general-purpose computer system.

[0076] In some embodiments, the computing environment 100 may also be a subsystem of one of the systems mentioned above. In some other embodiments, the computing environment 100 may be a “off-the-shelf” general-purpose computer system. In some embodiments, the computing environment 100 may also be distributed across multiple systems. The computing environment 100 may also be specific to an implementation of the Art. As those skilled in the Art will understand, several variations in how the computing environment 100 is implemented can be assumed without departing from the scope of the Art.

[0077] Those skilled in the art will understand that processor 110 generally represents processing power. In some embodiments, one or more dedicated processing cores may be provided instead of or in addition to one or more conventional central processing units (CPUs). For example, one or more graphics processing units 111 (GPUs), tensor processing units (TPUs), and / or other so-called acceleration processors (or processing accelerators) may be provided in addition to or in addition to one or more CPUs.

[0078] System memory typically includes random access memory 130, but more generally, it is intended to encompass any type of non-temporary system memory, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous DRAM (SDRAM), read-only memory (ROM), or a combination thereof. While a solid-state drive 120 is shown as an example of a mass storage device, more generally, such a mass storage device may comprise any type of non-temporary storage device configured to store data, programs, and other information and to make that information accessible via the system bus 160. For example, a mass storage device may comprise one or more of a solid-state drive, a hard disk drive, a magnetic disk drive, and / or an optical disk drive.

[0079] Communication between various components of the computing environment 100 may be enabled by a system bus 160 having one or more internal and / or external buses (e.g., PCI bus, Universal Serial Bus, IEEE 1394, "Firewire" bus, SCSI bus, Serial ATA bus, ARINC bus, etc.) to which various hardware components are electronically coupled.

[0080] The input / output interface 150 may enable networking capabilities, such as wired or wireless access. For example, the input / output interface 150 may include, but is not limited to, networking interfaces such as network ports, network sockets, and network interface controllers. Several examples of how networking interfaces may be implemented will become apparent to those skilled in the art. For instance, a networking interface may implement specific physical and data link layer standards, such as Ethernet, Fibre Channel, Wi-Fi, Token Ring, or serial communication protocols. These specific physical and data link layers may provide the foundation for a complete network protocol stack, enabling communication between small groups of computers on the same local area network (LAN) and large-scale network communication over routable protocols such as the Internet Protocol (IP).

[0081] The input / output interface 150 may be coupled to the touchscreen 190 and / or one or more internal and / or external buses 160. The touchscreen 190 may be part of the display. In some embodiments, the touchscreen 190 is the display. The touchscreen 190 may also be referred to as the screen 190. In the embodiment shown in Figure 1, the touchscreen 190 comprises touch hardware 194 (e.g., pressure-sensitive cells embedded within a layer of the display that enable detection of physical interaction between the user and the display) and a touch input / output controller 192 that enables communication with the display interface 140 and / or one or more internal and / or external buses 160. In some embodiments, the input / output interface 150 may be coupled to a keyboard (not shown), mouse (not shown), or trackpad (not shown) that allows the user to interact with a computing device 100 in addition to, or instead of, the touchscreen 190.

[0082] According to some implementations of this technology, the solid-state drive 120 stores program instructions suitable for being loaded into the random-access memory 130 and executed by the processor 110 in order to perform one or more of the actions described herein. For example, at least some of the program instructions may be part of a library or application.

[0083] Retinal signal processing system Figure 2 is a block diagram of the retinal signal processing system 200 according to various embodiments of the present technology. The retinal signal processing system 200 can collect retinal signal data from a patient. As described above, compared to conventional ERGs, the retinal signal data captured using the retinal signal processing system 200 may include additional features and / or data such as impedance, higher measurement frequencies, extended range of retinal light stimulation, and / or longer measurement times. The retinal signal processing system 200 can process and / or analyze the collected data. The retinal signal processing system 200 may output a predictive likelihood that the patient is in a given state, such as a medical condition.

[0084] It should be clearly understood that the illustrated system 200 is merely an exemplary implementation of the Art. Therefore, the following description is intended to be merely an illustrative example of the Art. This description is not intended to define the scope of the Art or to clarify its boundaries. In some cases, examples of modifications to system 200 that may be useful are also described below. This is done simply to aid understanding and again not to define the scope of the Art or to clarify its boundaries. These variations are not an exhaustive list, and as those skilled in the art will understand, other modifications are likely possible. Furthermore, where this is not done (i.e., examples of modifications are not shown), it should not be interpreted as modification being impossible, and / or that what is described is the only way to implement its elements. As those skilled in the art will understand, this is likely not the case. In addition, it should be understood that system 200 may, in some cases, provide simpler implementations of the Art, and in such cases, they are thus presented for the purpose of aiding understanding. As those skilled in the art will understand, various implementations of the Art may be more complex.

[0085] The retinal signal processing system 200 may include an optical stimulator 205, which may be an optical stimulator, for providing optical stimulation signals to the patient's retina. The retinal signal processing system 200 may include a sensor 210 for collecting electrical signals generated in response to optical stimulation. The retinal signal processing system 200 may include a data acquisition system 215, which may be a computing environment 100, for controlling the optical stimulator 205 and / or for collecting data measured by the sensor 210. For example, the optical stimulator 205 and / or the sensor 210 may be commercially available ERG systems, such as the Espion Visual Electrophysiology System from DIAGNOSYS, LLC, or the UTAS and RETEVAL systems manufactured by LKC TECHNOLOGIES, INC.

[0086] The photostimulator 205 may be any type of light source capable of generating light within a specified range of wavelength, intensity, frequency, and / or duration, either alone or in combination. The photostimulator 205 may direct the generated light onto the patient's retina. The photostimulator 205 may comprise light-emitting diodes (LEDs) in combination with one or more other light sources, such as xenon lamps. The photostimulator 205 may provide a background light source.

[0087] The optical stimulator 205 may be configured to provide an optical stimulation signal to the patient's retina. The collected retinal signal data may be dependent on the optical stimulation conditions. To maximize the potential for generating relevant retinal signal features in the retinal signal data, the optical stimulator 205 may be configured to provide a wide variety of optical conditions. The optical stimulator 205 may be configured to control the background light and / or the stimulation light directed onto the retina as a flash.

[0088] Light stimulators use different wavelengths (e.g., approximately 300 to 800 nanometers) and light intensities (e.g., 0.001 to approximately 3000 cd.s / m²). 2 ), irradiation time (e.g., approximately 1 to approximately 500 milliseconds), different background wavelengths (e.g., approximately 300 to approximately 800 nanometers) and background brightness (e.g., approximately 0.01 to approximately 900 cd / m²) 2 The system may include any light source capable of generating a light beam with a time between each flash (e.g., about 0.2 to 50 seconds) accompanied by a luminescence.

[0089] The retinal signal processing system 200 may include a sensor 210. The sensor 210 may be positioned to detect electrical signals from the retina. The sensor 210 may include one or more electrodes. The sensor 210 may be an electroretinogram sensor. A ground electrode may be positioned on the skin in the center of the forehead. Reference electrodes for each eye may be positioned on the earlobe, the temporal region near the eye, or other skin area.

[0090] Electrical signals from the retina can be triggered by light stimulation from the photostimulator 205 and collected as retinal signal data by the sensor 210. The retinal signal data can be collected by the sensor 210, such as electrodes placed on the eyeball or a nearby area of ​​the eyeball. Light can trigger low-amplitude electrical signals generated by the patient's retinal cells. Different types of retinal cells can be triggered, and therefore different electrical signals can be generated, depending on the properties of the light (e.g., intensity, wavelength, spectrum, frequency, and duration of the flash) and the conditions related to the light stimulation (e.g., background light, dark or light adaptation of the individual undergoing this process). This signal propagates within the eye and eventually through the optic nerve to the visual cortex of the brain. However, like any electrical signal, it propagates in all possible directions depending on the conductivity of the tissue. Therefore, the electrical signal can be collected in externally accessible tissues outside the eyeball, such as the conjunctiva.

[0091] Several types of electrodes exist that can be used to collect electrical signals. They are based on specific wire conductivity and geometric shape. It should be understood that many possible designs of recording electrodes exist, and any suitable design or combination of designs can be used for sensor 210. Sensor 210 may comprise contact lenses, gold foil, gold wire, corneal wick, wire loop, microfiber, and / or skin electrodes. Each electrode type has its own recording characteristics and inherent artifacts.

[0092] In addition to sensor 210, system 200 may also include other devices that monitor eye position and / or pupil size, both of which affect the amount of light reaching the retina and therefore affect the electrical signals triggered in response to this stimulus. System 200 may also include sensors that record light wavelength, light spectrum, and / or light intensity, such as a spectrometer and / or photodetector.

[0093] The electrical signal can be acquired between an active ocular electrode (placed above the eye) and a reference electrode by differential recording from a ground electrode. The electrodes of sensor 210 can be connected to a data acquisition system 215 which may include a recording device. The data acquisition system 215 may enable amplification of the electrical signal and / or conversion of the electrical signal to a digital signal for further processing. The data acquisition system 215 may implement a frequency filtering process that can be applied to the electrical signal from sensor 210. The data acquisition system 215 may store the electrical signal in a database in voltage-to-time format.

[0094] The data acquisition system 215 may be configured to receive patient measurement electrical signals from a sensor 210 and / or stimulating light data from an optical stimulator 205, and to store this acquired data as retinal signal data. The data acquisition system 215 may be operably coupled to an optical stimulator 205, which may be configured to trigger electrical signals and provide data to the data acquisition system 215. The data acquisition system 215 may synchronize the light stimulation with the acquisition and recording of electrical signals.

[0095] The collected data may be provided to the data acquisition system 215 via any suitable method, such as via a storage device (not shown) and / or a network. The data acquisition system 215 may be connectable to the sensor 210 and / or the optical stimulator 205 via a communication network (not shown). The communication network may be the Internet and / or an intranet. Multiple embodiments of the communication network may be conceivable and will be apparent to those skilled in the art.

[0096] Retinal signal data may include electrical response data (e.g., voltage and circuit impedance) collected over several signal acquisition times (e.g., 5 to 500 milliseconds) at several sampling frequencies (e.g., 0.2 to 24 kHz) with an optical stimulation synchronization time (flash duration) and / or offset (baseline voltage and impedance prior to the optical stimulation). The acquisition system 215 may collect retinal signal data at frequencies (i.e., sampling rates) of 4 to 16 kHz or higher. This frequency may be higher than that of conventional ERGs. Electrical response data may be collected continuously or intermittently.

[0097] The data acquisition system 215 may include a sensor processor for measuring the impedance of an electrical circuit used to acquire retinal signal data. The impedance of the electrical circuit may be recorded simultaneously with the acquisition of other electrical signals. The acquired impedance data may be stored within the retinal signal data. A method for determining the impedance of the circuit simultaneously with the acquisition of electrical signals may be based on a process of injecting a reference signal of known frequency and amplitude through the recording channel of the electrical signal. This reference signal may then be filtered and processed individually. The electrode impedance may be calculated by measuring the magnitude of the output at the excitation signal frequency. The impedance may then be used as a covariate to increase the signal density using the resistance value of the circuit at each point in time of recording the electrical signal.

[0098] The data analysis system 220 can process the data collected by the data acquisition system 215. The data analysis system 220 can extract retinal signal features and / or descriptors from the retinal signal data, and / or perform any other processing on the retinal signal data. The data analysis system 220 can receive the patient's clinical information and / or extract clinical information cofactors from the clinical information.

[0099] The predictive output system 225 receives data from the data analysis system 220 and can generate outputs to be used by clinicians. The outputs may be an output user interface, a report, or other documents. The outputs may show the predicted likelihood that a patient is in one or more conditions. For each condition, the outputs may show the predicted likelihood that the patient is in that condition. The outputs may show the patient's positioning within a pathology. The outputs may be used by clinicians to determine whether a patient is in a medical condition and / or to help determine what medical condition the patient is in.

[0100] The data collection system 215, the data analysis system 220, and / or the predictive output system 225 may be accessed by one or more users, such as through their respective clinics and / or through a server (not shown). The data collection system 215, the data analysis system 220, and / or the predictive output system 225 may also be connected to appointment management software that can schedule appointments or follow-ups based on status determinations by an embodiment of system 200.

[0101] The data acquisition system 215, the data analysis system 220, and / or the predictive output system 225 may be distributed across multiple systems and / or combined within one or more systems. The data acquisition system 215, the data analysis system 220, and / or the predictive output system 225 may be geographically distributed.

[0102] The systems and methods described herein may include: 1) retrieving retinal signal data collected from multiple individuals from the memory of a computer system, etc.; 2) extracting and / or generating retinal signal features from the retinal signal data, such as voltage, circuit impedance, signal acquisition time, sampling frequency, photostimulus synchronization time and / or offset, and / or any other type of data, which can be extracted from or generated using the retinal signal data; 3) combining the retinal signal features with cofactors from clinical information related to the observed state in each individual; 4) selecting retinal signal features from the extracted and combined features, ranking them according to their relevance, and determining a hierarchy; 5) assembling this information into mathematical descriptors; 6) inferring classifiers from those mathematical descriptors; 7) constructing mathematical domains of classifiers related to those states; and / or 8) obtaining density functions from those classifiers.

[0103] A mathematical descriptor can be a mathematical function that combines features from retinal signal data and / or clinical cofactors. The descriptor may exhibit retinal signal features specific to a state or population in terms of further differentiation between groups of patients. Examples of descriptors that can be used include skewness, kurtosis, compactness, eigenvectors, centroid coordinates, local binary patterns, time-series regression coefficients, spectral entropy, quantum entropy of any form, Rennie entropy, von Neumann entropy, Hartley entropy, Tsaris entropy, integrated entropy, Hu moment, Haralic features, and / or eigenvalue-based functions. Descriptors can be used to obtain classifiers. Classifiers can be mathematical or statistical functions that use descriptors to map data into categories or classes of information by ranking the descriptors according to their statistical significance. Descriptors can be ranked based on their relevance to depict specific components of a biosignature. Descriptors can be grouped into a descriptor catalog. A classifier catalog can be used when training an MLA.

[0104] The systems and methods described herein may include: 1) retrieving retinal signal data collected from an individual from the memory of a computer system or the like; 2) extracting and / or generating retinal signal features from the retinal signal data, such as voltage, circuit impedance, signal acquisition time, sampling frequency, photostimulus synchronization time and / or offset, and / or any other type of data, which can be extracted from or generated using the retinal signal data; 3) combining the retinal signal features with cofactors from clinical information related to the condition observed in the individual; 4) calculating the probability that the individual belongs to one or more domains (i.e., is in a condition); and / or 5) determining the mathematical proximity of the individual to those domains as a predictive probability that the patient is in the condition corresponding to the domain.

[0105] Clinical information may include information relating to an individual's general health status, such as information on comorbidities, treatments, previous medical conditions, coffee, alcohol, or tobacco use, substance abuse, and / or any other general health status data. Clinical information may include information relating to a specific psychiatric condition, such as information from structured questionnaires specific to mental disorders. These structured questionnaires may include questions relating to anxiety, feelings, and mood components, cognitive impairment, emotions, habits, hallucinations, behavior, and / or other questions relating to mental disorders. Clinical information adjuncts, such as clinical information adjuncts indicating age, sex, iris color, and / or skin pigmentation as a surrogate for retinal pigmentation, may be extracted from clinical information.

[0106] Retinal signal data may contain several biosignatures, each biosignature being condition-specific. Biosignatures corresponding to a given patient's condition can be identified using a classifier. Conditions may be, but are not limited to, psychiatric conditions such as bipolar disorder, schizophrenia, and depression. Conditions may also be neurological conditions, non-psychiatric conditions, or conditions at risk of such conditions. The steps for analyzing retinal signal data described herein may be applicable to the analysis of retinal signal features specific to any condition represented in the retinal signal data.

[0107] A graphical representation of the findings can be created and output, providing visual support for the selections made when choosing retinal signal features to be used in mathematical models. The application can apply mathematical and / or statistical analysis of the results to evaluate the fit of the data generated during the analysis process, the robustness of the information, and the accuracy of the results.

[0108] Retinal signal features extracted from recorded retinal signal data may include electrical parameters such as voltage and circuit impedance, signal acquisition time (e.g., 5 to 500 milliseconds), sampling frequency (e.g., 0.2 to 24 kHz), light stimulation synchronization time (flash duration) and offset (baseline voltage and impedance before light stimulation), illuminated retinal area, and / or other retinal signal features that affect the retinal signal data. Retinal signal features may be generated based on the extracted retinal signal features, for example, by performing mathematical operations on one or more of the extracted retinal signal features. Retinal signal features extracted from retinal signal data may include data related to the retinal signal, such as eye position, pupil size, distance from the light source to the eye or part of the eye (pupil, retina), and / or applied luminance parameters (intensity, wavelength, frequency of retinal signal sampling, wavelength, illumination time, background wavelength, background luminance). Retinal signal features may be voltage-independent and / or time-independent. All or some of these features may be analyzed using the systems and methods described herein.

[0109] Retinal signal data processing method Figure 3 shows an exemplary functional architecture of information processing method 300, which leads to the construction of a mathematical function for predicting whether a patient is in a state. Method 300 can be used to model the mathematical domain of information from retinal signal data collected from patients in a state. All or part of Method 300 may be performed during the information input stage 310, the feature extraction stage 320, the feature selection and positioning stage 330, the feature weighting and assembly stage 340, the classifier ranking and distinction stage 350, the face stage 360, and / or the predictive output stage 370.

[0110] Retinal signal data and collected clinical information may be processed differently depending on the level and specificity of the information. Retinal signal features may be classes or categories of indices given by explanatory variables directed toward attributes. Descriptors may be relevant features specific to a state or population, determined according to a distinction process. Classifiers may be mathematical or statistical functions that use descriptors to map data into categories or classes according to a ranking process. Regions may be identified as a subset of a range of mathematical functions or collections of functions constructed using classifiers. Domains may be identified as regions specific to a state.

[0111] It should be clearly understood that the functional architecture shown in Figure 3 is merely an exemplary implementation of this technology. Therefore, the following description is intended to be merely an illustrative example of this technology. This description is not intended to define the scope of this technology or to clarify its boundaries.

[0112] In some cases, examples of modifications to the functional architecture shown in Figure 3 may be described below, which may be considered useful examples of modifications to the functional architecture shown in Figure 3. This is done simply to aid understanding and again not to define the scope of the Art or to clarify its boundaries. These variations are not an exhaustive list, and as those skilled in the art will understand, other modifications are likely possible. Furthermore, if this is not done (i.e., no examples of modifications are shown), it should not be interpreted as modifications being impossible, and / or that what is described is the only way to implement that element of the Art. As those skilled in the art will understand, this is likely not the case. In addition, the functional architecture shown in Figure 3 may, in some cases, provide a simpler implementation of the Art, and in such cases, it should be understood that they are presented in this manner for the purpose of aiding understanding. As those skilled in the art will understand, various implementations of the Art may be more complex.

[0113] In one or more embodiments, Method 300 or one or more steps thereof may be performed by a computing system such as a computing environment 100. Method 300 or one or more steps thereof may be embodied in computer executable instructions stored in a computer-readable medium such as a non-temporary mass storage device, loaded into memory, and executed by a CPU. It should be understood that Method 300 is illustrative and that some steps or parts of steps in the figures may be omitted and / or the order may be changed.

[0114] In certain embodiments, the possibility of specific mathematical modeling of a state is considered during the demonstration and subsequently supported by a graphical representation of the state-specific information domain used to explain the concept of the mathematical domain. The results of robustness and accuracy analyses are used to demonstrate the relevance and specificity of the mathematical domain constructed using this technique.

[0115] Methods for deciphering biosignatures contained within retinal signal data can be enhanced by extending the analytical policy to additional computational processes and subsequently generating additional mathematical descriptors.

[0116] Information entry In step 311, information input may be performed using the collected retinal signal data with or without removing artifacts, which may include distorted signals, interference, and / or any other type of artifacts. Artifacts may be caused by the inadvertent capture of electrical signals not originating from the retina, shifts in the positioning of the ocular electrodes, changes in the contact of the ground or reference electrodes, blinking of the eyelids, and / or eye movements.

[0117] In step 312, the collected retinal signal data may be transformed by, for example, scaling, shifting, elementary function transformations (e.g., logarithmic, polynomial, power, trigonometric functions), time series (resulting in a change in distribution shape), wavelet transform (resulting in a scalogram), empirical mode decomposition (resulting in an eigenmode function (IMF)), gradient transform (resulting in a vector), and / or kernel decomposition (resulting in a change in distribution shape and / or length scale as well as signal dispersion), such as transposing the retinal signal data to another one-dimensional or multidimensional scale. Filtering such as high-pass filters, low-pass filters, band-pass filters, and notch filters (finite impulse response, infinite impulse response) may be performed on the retinal signal data. Hilbert fan transform (instantaneous frequency of IMF).

[0118] In step 313, the clinical information may be transformed by, for example, scaling, shifting, elementary function transformations (e.g., logarithmic, polynomial, power, trigonometric functions), and / or regrouping of variables in a compound variable, such as transposing the clinical data to another one-dimensional or multidimensional scale.

[0119] In step 314, the data conformity of the data generated during the information input stage 310 can be tested and verified by ensuring that all data to be processed conforms to a standardized format suitable for the intended processing.

[0120] Feature extraction In step 321, retinal signal features that are considered to contain components of the biosignature may be obtained by, for example, time-frequency analysis (magnitude, minimum and maximum magnitude locations at specific points in the scalogram), kernel decomposition, principal component analysis (PCA), geometric (algebraic) operations at various time-frequency intervals such as minimum (e.g., wave a), maximum (e.g., wave b), latency, slope, gradient, curvature, integral, energy (sum of squares of amplitude), variance, cohesive-variance (uniformity, density), and / or any other method for extracting or generating retinal signal features from retinal signal data.

[0121] In step 322, potential clinical information cofactors that are considered to influence the components of the biosignature may be obtained from clinical information by, for example, multicomponent analysis (chi-squared combined with variance) and subsequent grouping, forward selection based on stepwise regression, best subset regression (using a specified set of cofactors), from clinical practice (i.e., cofactors for a condition, e.g., duration of disease, number of crises / hospitalizations), and / or any other method for obtaining clinical information cofactors from clinical information.

[0122] In step 323, the robustness of the information generated during the feature extraction step 320 and the accuracy of the results can be evaluated.

[0123] Feature selection and positioning In step 331, the retinal signal features containing the most important components of the biosignature may be selected by, for example, time-frequency visual analysis or scalogram (in which case the retinal signal features are the magnitude of the signal in a particular time-frequency window where distinctions are found), stepwise regression with cross-validation, cofactor-adjusted or unadjusted sparse representation-based classification (SRC) with predefined thresholds, selection and combination of the most relevant retinal signal features to generate principal components in a supervised PCA (SPCA) process, least absolute contraction and selection operator (LASSO), ridge regression, elastic net, Bayesian or spike-and-slab method, and / or any other selection method.

[0124] In step 332, for example, the cofactor hierarchy (confounding, mediators, moderators) may be determined by i) confounding methods (randomization, restriction, and matching), stratification and subsequent Mantel-Henzel estimators, and / or multivariate methods, e.g., ANCOVA, linear regression, logistic regression, and / or ii) mediation and adjustment methods, e.g., the Baron and Kenny method, the Fairchild and MacKinnon method, and / or other appropriate methods, that is, the direction and hierarchy of the influence of the most important retinal signal features and clinical information cofactors as components of the biosignature, which together have a high contribution to the analytical model.

[0125] In step 333, clinical information cofactors that have a high contribution to the mathematical model, i.e., clinical information cofactors that influence the components of the biosignature and together have a high contribution to the mathematical model (and therefore become "eligible" cofactors), are selected from the clinical information.

[0126] In step 334, the robustness of the information generated during the feature selection and positioning stage 330 and the accuracy of the results can be evaluated.

[0127] Feature weighting and assembly In step 341, the relevant retinal signal features may be reconstructed and / or weighted to generate descriptors. For example, the contributions of retinal signal features may be determined and / or mapped. Retinal signal features that include biosignature descriptors and have an impact on the model may be identified, for example, by multivariable regression analysis or related processes.

[0128] In step 342, descriptors may be assembled using the most important clinical information cofactors (i.e., eligible cofactors). Then, descriptors may be selected to obtain the biosignature components that together contribute most to the mathematical model, for example, by matching and merging descriptors and cofactors using formulas or relations, for example, by using other methods used in selecting and / or combining PCA, SPCA, or retinal signal data features.

[0129] In step 343, the robustness of the information generated during the feature weighting and assembly stages 340 and the accuracy of the results can be evaluated.

[0130] Classifier ranking and distinction In step 351, classifiers may be estimated by using assembled descriptors and / or clinical information cofactors to train the classifiers. Classifiers may be selected and / or ranked based on their performance by, for example, logistic regression, probit regression, stochastic modeling (Gaussian processes, kernel estimation models, Bayesian modeling), best subset regression (using a specified set of predictors), SVM, neural network methods, decision trees, random forests, weighted voting, boosting and bagging, Kaplan-Meier analysis, Cox regression, and / or other selection or ranking methods.

[0131] In step 352, regions and domains can be constructed by selecting state-specific classifiers and mapping them to mathematical functions that represent their most distinctive biosignatures, for example, by visualization of regression results enhanced with mathematical constructs (e.g., wavelets, Kullback-Leibler divergence, etc.), Neyman-Pearson for distinguishing between domains, and / or other relevant methods.

[0132] In step 353, the robustness of the information generated during the classifier ranking and distinction stage 350, as well as the accuracy of the results, can be evaluated.

[0133] We face findings from retinal signal decoding. In step 361, density functions may be obtained and compared, i.e., by obtaining a mathematical function containing the highest density of biosignature components from retinal signal data and clinical cofactors, and by comparing those functions across states, for example, by histograms, kernel density estimation (e.g., Partzen-Rosenblatt window, bandwidth selection), characteristic function density and other relevant estimators, data clustering techniques including vector quantization, reconstruction methods based on either sample cumulative probability distributions or sample moments, and / or other methods.

[0134] Complete the decoding of retinal signals. In step 371, when comparing data obtained under various conditions, probabilities can be calculated from a mathematical function that includes the relevant density of biosignature components from retinal signal data and clinical information cofactors, for example, by using a density function and estimating the CPF (cumulative probability function), and / or by other relevant methods.

[0135] In step 372, the distance from a region can be determined by identifying the biosignature component most relevant to the distance between regions or domains with high information density in various states, and by calculating the distance as a formula, for example by comparing the probability with a Bayesian statistical prior probability; entropy-based methods; the Kullback-Leibler divergence parameter for assessing the direction of change; and the Neyman-Pearson for assessing the distance between regions. The semiparametric maximum likelihood estimation procedure Pooled Adjacent Violation Algorithm (PAVA) can be used, as well as other methods that challenge the medical states to which individual parts of the model (i.e., regions) belong and reassess their attribution to the model using IPWE (Inverse Probability Weighted Estimator) or related methods.

[0136] In step 373, the robustness of the information generated during the completion stage 370 and the accuracy of the results can be evaluated.

[0137] Prediction method Figure 4 shows a flowchart of Method 400 for predicting the likelihood of a patient being in a condition, according to various embodiments of the technique. All or part of Method 400 may be performed by a data acquisition system 215, a data analysis system 220, and / or a predictive output system 225. In one or more embodiments, Method 400 or one or more steps thereof may be performed by a computing system, such as a computing environment 100. Method 400 or one or more steps thereof may be embodied in computer-executable instructions stored in a computer-readable medium, such as a non-temporary mass storage device, loaded into memory, and executed by a CPU. It should be understood that Method 400 is illustrative and that some steps or parts of steps in the flowchart may be omitted and / or their order may be changed.

[0138] Method 400 includes steps that perform various actions, as will be described in more detail below, such as extracting retinal signal features from retinal signal data, selecting retinal signal features most relevant to a particular state, combining and comparing those retinal features to generate mathematical descriptors that best distinguish the states to be analyzed or compared, generating multimodal mappings, identifying biomarkers and / or biosignatures of the states, and / or predicting the likelihood that a patient is in any one of the states.

[0139] In step 405, retinal signal data may be acquired from the patient. Retinal signal data may be acquired using a predefined acquisition protocol. Retinal signal data may include measured electrical signals acquired by electrodes placed on the patient. Retinal signal data may include parameters of the system used to acquire retinal signal data, such as parameters of the light stimulus. Retinal signal data may include the impedance of the receiving electrical circuit used in the device that measures the electrical signals. In certain embodiments, this step 405 is omitted.

[0140] Retinal signal data may include impedance measurements and / or other electrical parameters. Retinal signal data may also include optical parameters such as changes in pupil size and / or applicable luminance parameters (intensity, wavelength, spectrum, frequency of light stimulation, frequency of retinal signal sampling).

[0141] To generate retinal signal data, the patient's retina may be stimulated, for example, by using an optical stimulator 205, which may be one or more optical stimulants. The retinal signal data may be collected by a sensor, such as a sensor 210, which may comprise one or more electrodes and / or other sensors.

[0142] The light stimulator uses different wavelengths (e.g., approximately 300 to 800 nanometers) and light intensities (e.g., approximately 0.01 to 3000 cd.s / m²). 2), irradiation time (e.g., approximately 1 to approximately 500 milliseconds), different background wavelengths (e.g., approximately 300 to approximately 800 nanometers) and background brightness (e.g., approximately 0.1 to approximately 800 cd / m²) 2 The system may include any light source capable of generating a light beam with a time interval between each flash (e.g., about 0.2 to about 50 seconds) accompanied by a light source.

[0143] Retinal signal data may include electrical response data (e.g., voltage and circuit impedance) collected over several signal acquisition times (e.g., 5 to 500 milliseconds) at several sampling frequencies (e.g., 0.2 to 24 kHz), with a light stimulation synchronization time (flash duration) and offset (baseline voltage and impedance prior to the light stimulation). Therefore, step 405 may include a step of collecting retinal signal data at frequencies from 4 to 16 kHz.

[0144] For example, after retinal signal data is collected by a practicing physician, the retinal signal data can be uploaded to a server such as a data analysis system 220 for analysis. The retinal signal data can be stored in the memory 130 of a computer system.

[0145] If step 405 is omitted in step 410, the retinal signal data may be retrieved from memory 130. Retinal signal features may be extracted from the retinal signal data. Extraction of retinal signal features may be based on processing the retinal signal data and / or their transformations using multiple signal analysis methods such as polynomial regression, wavelet transform, and / or empirical mode decomposition (EMD). Extraction of retinal signal features may be based on parameters derived from their analysis or specific modeling, e.g., principal components and maximum distinction predictors, parameters from linear or nonlinear regression functions, higher magnitude frequencies, Kullback-Leibler coefficients of differences, Gaussian kernel features, log-likelihoods of differences, and / or high-energy regions. These analyses may be used to determine the contribution of each specific retinal signal feature and to statistically compare the retinal signal features.

[0146] The retinal signal features to be extracted may have been determined previously. The retinal signal features to be extracted may have been determined by analyzing labeled datasets of retinal signal data from multiple patients. Each patient represented in the dataset may have one or more associated medical conditions, and / or one or more medical conditions that the patient does not have. These medical conditions may be labels for each patient's dataset. The retinal signal features to be extracted may be determined by analyzing sets of retinal signal data from patients who share the same medical conditions. Based on the retinal signal features, a multimodal map may be generated. Based on the multimodal map, a domain may be determined.

[0147] In step 415, descriptors may be extracted from retinal signal features. Mathematical descriptors may be mathematical functions that combine features from retinal signal data and / or clinical cofactors. Descriptors may exhibit retinal signal features specific to a state or population in terms of further differentiation between groups of patients. Descriptors may be selected to obtain biosignature components that contribute most to the mathematical model together, for example, by using PCA, SPCA, or other methods used in selecting and / or combining retinal signal data features, as described above in step 342 of Method 300, or by matching and merging descriptors and cofactors using mathematical formulas or relationships.

[0148] In step 420, patient clinical information may be received. Clinical information may include medical records and / or any other data collected about the patient. Clinical data may include questionnaires taken by healthcare professionals and / or results of clinical tests.

[0149] In step 425, clinical information may be used to generate clinical information cofactors. Clinical information cofactors may be selected based on their influence on retinal signal data. Clinical information cofactors may include the patient's age, sex, skin pigmentation which may be used as a surrogate for retinal pigmentation, and / or any other indicators of clinical information relevant to the patient.

[0150] In step 430, clinical information cofactors and / or descriptors may be applied to mathematical models of a state. Any number of mathematical models may be used. The clinician may choose which mathematical models to use. Each model may correspond to a specific state or control.

[0151] In step 435, each model may determine the distance between the patient and the model's biosignature. The main components of the retinal signal data may be located within the domain corresponding to the state. Descriptors and / or clinical information cofactors may be compared to the biosignature of each model.

[0152] In step 440, each model may output a predicted probability that the patient is in the model state. The likelihood of the patient being in the state may be predicted based on the level of statistical significance when comparing the size and location of the individual's descriptor with those in the model. The predicted probability may be binary and may indicate whether or not the biosignature of the state is present in the patient's retinal signal data. The predicted probability may be a percentage indicating how likely the patient is to be in the state.

[0153] In step 445, the predicted probability of the patient being in each state may be output. An interface and / or report may be output. The interface may be output on a display. The interface and / or report may be output to the clinician. The output may show the likelihood that the patient is in one or more states. The output may show the patient's positioning within the pathology. The predicted probabilities may be stored.

[0154] The output may include determining the medical condition, the predicted probability of the medical condition, and / or the degree to which the patient's retinal signal data matches the condition and / or other conditions. The predicted probability may be in the form of a percentage of correspondence with the medical condition, which may provide an objective neurophysiological measure to further support the clinician's hypothesis of the medical condition.

[0155] The output may be used in conjunction with a clinician's tentative medical condition hypothesis to increase the clinician's confidence in determining the medical condition and / or to initiate an earlier or more effective treatment plan. The output may be used to initiate treatment earlier rather than spending additional time clarifying the medical condition and treatment plan. The output may reduce the level of uncertainty for the clinician and / or patient regarding the clinician's tentative medical condition hypothesis. The output may be used to select a drug to administer to the patient. The selected drug may then be administered to the patient.

[0156] Method 400 may be used to monitor a patient's condition. The patient may have been previously diagnosed with a condition. Method 400 may be used to monitor the progression of a condition. Method 400 may be used to monitor and / or modify a treatment plan for a condition. For example, Method 400 may be used to monitor the effectiveness of medications being used to treat a condition. Retinal signal data may be collected before, during, and / or after the patient receives treatment for a condition.

[0157] Method 400 may be used to identify and / or monitor neurological symptoms of infectious diseases, such as viral infections. For example, Method 400 may be used to identify and / or monitor neurological symptoms in patients infected with COVID-19. Retinal signal data may be collected from patients who are infected with or have been infected with COVID-19. Retinal signal data may be evaluated using Method 400 to determine whether the patient has neurological symptoms, the severity of the neurological symptoms, and / or to develop a treatment plan for the neurological symptoms.

[0158] Generate mathematical models Figure 5 shows a flowchart of Method 500 for generating a mathematical model to predict whether a patient is in a certain condition, according to various embodiments of the present technology. All or part of Method 500 may be performed by a data acquisition system 215, a data analysis system 220, and / or a prediction output system 225. In one or more embodiments, Method 500 or one or more steps thereof may be performed by a computing system, such as a computing environment 100. Method 500 or one or more steps thereof may be embodied in computer executable instructions that are stored in a computer-readable medium, such as a non-temporary mass storage device, loaded into memory, and executed by a CPU. It should be understood that Method 500 is illustrative and that some or some steps in the flowchart may be omitted and / or the order may be changed.

[0159] Method 500 may process datasets of retinal signal data and / or other clinical information to create classification models based on domains specific to various states. Mathematical modeling of retinal signal data may be performed by using several sequential analyses that process retinal signal data and / or clinical information cofactors to generate classification metrics based on domains specific to states. The analyses have several mechanisms that combine retinal signal data with clinical information cofactors, such as specific descriptors within retinal biosignatures, to select the most distinguishable components as descriptors. These descriptors, which may be combined with clinical information cofactors, may be used to train classifiers and then selected based on their performance in providing probability factors that a patient is in a state. Descriptors may be used to construct density functions specific to each of the states. Domain clusters may be formed using multimodal analysis of distinguishing features and mappings of principal components and statistically significant descriptors.

[0160] In step 505, retinal signal data may be collected from the patient. The collected retinal signal data may include impedance measurements and / or other electrical parameters. The retinal signal data may include optical parameters such as the distance from the light source to the eye or part of the eye (pupil, retina), changes in pupil size, and / or applicable luminance parameters (light intensity, light wavelength, light spectrum, frequency of light stimulation, frequency of retinal signal sampling). In certain embodiments, step 505 may be omitted.

[0161] In step 510, each patient's retinal signal data may be labeled with a condition. The label may be one or more conditions to which the patient has been diagnosed. The label may indicate that the patient has not been diagnosed with any condition, in which case the patient may be referred to as a control. Conditions may include schizophrenia, bipolar disorder, major depressive disorder, psychosis, post-traumatic stress disorder, stroke, substance abuse, obsessive-compulsive disorder, Alzheimer's disease, Parkinson's disease, multiple sclerosis, autism, attention deficit disorder, and / or any other condition. If step 505 is omitted, step 510 may include retrieving the stored retinal signal data from memory, such as memory 130 of the computing environment 100.

[0162] In step 515, clinical information adjuncts regarding the patient may be received. Clinical information adjuncts may be extracted from medical records. Clinical information adjuncts may include indicators of population parameters such as age, sex, and / or skin pigmentation as a surrogate for retinal pigmentation, iris color, etc. Clinical information adjuncts may be received for all or a subset of patients from whom retinal signal data have been collected.

[0163] In step 520, a dataset may be generated for each patient. The dataset may include the patient's retinal signal data, labels assigned to the patient, and / or cofactors of the patient's clinical information. The dataset may be stored in a database and / or in any other suitable format.

[0164] In step 525, retinal signal features can be extracted from retinal signal data in the database. Any appropriate mathematical selection process, such as principal component analysis (PCA), generalized linear modeling (GLM), sparse representation-based classification (SRC), least absolute contraction and selection operator (LASSO), and / or a combination of several models, can be used to select the retinal features to extract, with the aim of selecting the most distinguishable retinal signal features, transforming the selected retinal signal features, and generating descriptors of interest in terms of cohesion and density (information density). A variety of methods, but not limited to, time-frequency analysis, prototype mother wavelets generated from a reference dataset, stochastic modeling, information cohesion and variance, and principal component analysis (PCA), generalized linear modeling (GLM), sparse representation-based classification (SRC), least absolute contraction and selection operator (LASSO), and / or similar selection methods or combinations of several methods, can be used to analyze the collected retinal signal data. Mapping can be used to depict the region of retinal signal data (n-dimensional) with the highest discriminative power in order to isolate the domains of each state and place individuals within those domains (distances).

[0165] A state's biosignature may include portions of retinal signal data specific to that state. One or more of the most distinctive retinal signal features may be selected as descriptors for a particular state. These descriptors may be combined in a multimodal computation process.

[0166] The possibility of specific mathematical modeling is explored during the demonstration and subsequently supported by a graphical representation of the domain of state-specific information, which was used to describe the domain. Among the retinal signal features obtained using multimodal analysis, clusters of relevant predictors can be defined.

[0167] By extending the analysis strategy to additional computational processes and subsequently generating additional retinal signal features, those components used to decode the biosignatures contained within retinal signal data can be enhanced.

[0168] Frequency (spectral) analysis Time-frequency analysis can be performed, for example, using a mother wavelet for discrete wavelet transform (DWT) or continuous wavelet transform (CWT) and / or empirical mode decomposition (EMD) filtering. Commonly used features, such as higher magnitude frequencies, occurrence times, frequency mapping (scalogram), time-frequency domains, wavelet coefficients as the output of the wavelet transform, relative occurrence frequencies of wavelet coefficients, and / or composite parameters, can be visualized and / or calculated.

[0169] A specific mother wavelet (prototype wavelet) can be constructed and tailored to the nature of the data (either a control group or a group of individuals with a particular pathology). The specific mother wavelet can be validated by reprocessing and filtering frequencies that are not considered important features (wavelet design). This strategy can be used to tune known, predefined mother wavelets and perform more specific time-frequency analysis with or without time clustering (i.e., time-frequency analysis with mother wavelets tailored according to the portion of the retinal signal waveform being analyzed).

[0170] The application is used to compare analyses and show differences between retinal signal data from patients with conditions such as psychiatric states, compared to retinal signal data from individuals not diagnosed with those conditions (control subjects).

[0171] Principal component analysis Principal component analysis (PCA) may be performed. PCA can be supervised (SPCA) or unsupervised (PCA) to extract the most distinguishable retinal signal features from retinal signal data. PCA can determine so-called "principal components," which may be the retinal signal features with the greatest dependence on the response variable, in this case the suspected state. PCA can analyze the dependency relationships between retinal signal features and states, enabling the selection of the retinal signal features with the greatest dependency.

[0172] Retinal signal features can be analyzed in combination with additional demographic or clinical variables specific to any associated pathology or medical condition. These analyses may yield specific data vectors (principal component scores) used in defining retinal signal features, which can then be combined and re-analyzed ("face-to-face") to derive a classification of the most distinct retinal signal features based on their statistical significance against predefined thresholds. Such classifications may allow for the selection of retinal signal features that can be used to construct vectorized domains specific to the condition.

[0173] Probabilistic modeling Probabilistic modeling may be used to find regions of interest within retinal signal data. Standard Gaussian and / or non-Gaussian process modeling may be used to define the probability of events within the large amount of collected information. The retinal signal features selected to build (train) the model may be chosen based on a dataset specific to retinal signal data from patients with a particular condition and a control set (a dataset from patients without the condition).

[0174] In step 530, a subset of retinal signal features corresponding to biomarkers of the condition may be selected. The retinal signal features may be ranked based on their relevance to the condition. Some or all of the highest-ranked retinal signal features may be selected so that they fall within the subset.

[0175] In step 535, a classifier may be determined to distinguish the biosignature of a state from the biosignatures of other states. The classifier may include a subset of retinal signal features and / or co-clinical information factors. The classifier may be a mathematical model that represents the biosignature. To predict the likelihood that a patient is in a state, the distance between the patient's retinal signal data and the state classifier may be determined.

[0176] Figure 6 shows a flowchart of Method 600 for training a machine learning algorithm (MLA) to predict the likelihood of a medical condition, such as a psychiatric condition or neurological symptoms, according to various embodiments of the present technology. All or part of Method 600 may be performed by a data acquisition system 215, a data analysis system 220, and / or a prediction output system 225. In one or more embodiments, Method 600 or one or more steps thereof may be performed by a computing system, such as a computing environment 100. Method 600 or one or more steps thereof may be embodied in computer executable instructions that are stored in a computer-readable medium, such as a non-temporary mass storage device, loaded into memory, and executed by a CPU. It should be understood that Method 600 is illustrative and that some steps or parts of steps in the flowchart may be omitted and / or the order may be changed.

[0177] In step 605, a dataset of retinal signal data may be extracted. The dataset may have been generated using steps 505-520 of method 500, as described above. Each dataset may include patient retinal signal data, one or more labels corresponding to the condition the patient was diagnosed with, and patient-related clinical informational adjuncts. The retinal signal data may correspond to multiple patients. The retinal signal data may be selected based on a specific patient population. For example, all retinal signal data corresponding to a specific sex and age range may be extracted.

[0178] The extracted dataset may include impedance measurements and / or other electrical parameters. The extracted dataset may include optical parameters such as changes in pupil size and / or applied luminance parameters (light intensity, wavelength, spectrum, frequency of light stimulation, frequency of retinal signal sampling). The extracted dataset may include population parameters such as age, sex, iris color, and / or skin pigmentation as a surrogate for retinal pigmentation.

[0179] Retinal signal data may be labeled. For each patient represented in the data, one or more conditions to which the patient has been diagnosed may be indicated. For patients not diagnosed with any of the available conditions, the condition may not be indicated, or a label indicating that the patient is a control may be indicated.

[0180] In step 610, retinal signal features may be generated. Retinal signal features may be generated using a transformation process. Retinal signal features may be generated for each set of retinal signal data extracted in step 605.

[0181] In step 615, descriptors may be extracted from retinal signal features within the dataset. These descriptors may be features selected as best representing a state.

[0182] In step 620, descriptors can be ranked. Each descriptor can be ranked based on its level of statistical significance when testing its contribution to distinguishing between states and / or between states and controls (i.e., undiagnosed states).

[0183] In step 625, descriptors may be selected based on their rankings. A predetermined number of the highest-ranked descriptors may be selected.

[0184] In step 630, the MLA may be trained using the selected descriptors. The dataset retrieved in step 605 may be filtered to remove descriptors other than those selected in step 625. Then, all or part of the dataset may be used to train the MLA. The MLA may be trained to predict the likelihood that a patient is in a state, based on the set of descriptors corresponding to the patient.

[0185] After training the MLA, it can be used to predict the likelihood that a patient is in a state based on measured retinal signal data. Figure 7 shows a flowchart of Method 700 for using MLA to predict the likelihood of a psychiatric state according to various embodiments of the present technology. All or part of Method 700 may be performed by a data acquisition system 215, a data analysis system 220, and / or a prediction output system 225. In one or more embodiments, Method 700 or one or more steps thereof may be performed by a computing system such as a computing environment 100. Method 700 or one or more steps thereof may be embodied in computer executable instructions that are stored in a computer-readable medium such as a non-temporary mass storage device, loaded into memory, and executed by a CPU. It should be understood that Method 700 is illustrative and that some steps or parts of steps in the flowchart may be omitted and / or the order may be changed.

[0186] In step 705, the patient's retinal signal data may be collected. The actions performed in step 505 may be the same as those described above with respect to step 405.

[0187] In step 710, retinal signal features and / or descriptors may be extracted from the retinal signal data. The descriptors may be extracted to correspond to an MLA, such as an MLA generated using method 600.

[0188] In step 715, descriptors may be entered into the MLA. Patient clinical information cofactors may also be entered into the MLA. The MLA may be configured to predict the likelihood that the patient is in a state based on the descriptors and / or clinical information cofactors. The MLA may be an MLA generated using method 600.

[0189] In step 720, a determination may be made as to whether further data should be input to the MLA. The MLA may determine that the data input in step 715 is insufficient to make a prediction. The MLA may determine that the amount of data is insufficient. The MLA may determine that the data is not accurate enough to make a prediction, for example, if errors are present in the data. The MLA may output a confidence level for the prediction, and it may be determined that the confidence level is below a threshold confidence level, such as a predetermined threshold confidence level. If the MLA requires more data, more data may be captured in step 705. If, instead, the MLA has enough data to make a prediction, method 700 may proceed to step 725.

[0190] In step 725, the MLA may output the predicted likelihood that the patient is in one or more conditions. For each condition, the MLA may output the predicted likelihood that the patient is in that condition. The patient's positioning within the pathology may be output by the MLA and / or determined based on the predictions output by the MLA. The patient's positioning may indicate the patient's distance from the biosignature of the condition. A user interface and / or report may be generated and output to the clinician for use as an aid in determining and / or confirming the patient's medical condition.

[0191] Examples of time-frequency analysis and selection of discriminant regions based on the statistical significance of higher-magnitude frequencies are shown in Figures 8 to 27. In these examples, retinal signal data from patients with schizophrenia, bipolar disorder, or major depressive disorder, young people at risk of psychosis (i.e., young offspring of parents with schizophrenia, bipolar disorder, or depression), and unaffected controls underwent time-frequency analysis (wavelet transform using different mother wavelets) and were compared. Time-frequency analysis is performed using any known mother wavelet, or a specific mother wavelet (prototype mother wavelet) prepared from a reference dataset (controls or individuals with a particular condition).

[0192] Examples of retinal signal feature selection and statistical significance mapping are shown in Figures 28 to 42. In these examples, retinal signal data obtained by signal-frequency analysis were selected and compared from patients with schizophrenia, bipolar disorder, or major depressive disorder, young people at risk of psychosis (i.e., young offspring of parents with schizophrenia, bipolar disorder, or depression), and control groups unaffected by these conditions. Mapping of the most distinguishing features is then prepared from the feature selection.

[0193] Examples of retinal signal descriptor selection and statistical significance mapping are shown in Figures 43 to 58. In these examples, retinal signal descriptors obtained from patients with schizophrenia, bipolar disorder, or major depressive disorder, young people at risk of psychosis (i.e., young offspring of parents with schizophrenia, bipolar disorder, or depression), and unaffected control subjects were combined, and their likelihood of differentiation within several conditions was evaluated. From this selection, a mapping of the most distinguishable descriptors is then prepared based on their distinguishability and their position within the volume of information.

[0194] Figure 8 shows a comparison of frequency analysis of retinal signal data in patients with schizophrenia (bright dots) and individuals not diagnosed with a mental health condition (control subjects, dark dots), using a discrete approximation of the Moret waveform as the mother wavelet. The magnitude of the SRC coefficient against time and frequency is a 90% threshold comparison.

[0195] Figure 9 shows a comparison of frequency analysis of retinal signal data in patients with schizophrenia (bright dots) and individuals not diagnosed with a mental health condition (controls, dark dots), using a discrete approximation of a mother wavelet specifically designed with retinal signal data sets from control subjects. The magnitude of the SRC coefficients against time and frequency is a 90% threshold comparison.

[0196] Figure 10 shows a comparison of frequency analysis of retinal signal data in patients with bipolar disorder (bright dots) and individuals not diagnosed with mental health status (control subjects, dark dots), using a discrete approximation of the Moret waveform as the mother wavelet. The magnitude of the SRC coefficient against time and frequency is a 90% threshold comparison.

[0197] Figure 11 shows a comparison of frequency analysis of retinal signal data in patients with bipolar disorder (bright dots) and individuals not diagnosed with a mental health condition (controls, dark dots), using a discrete approximation of a mother wavelet specifically designed with retinal signal data sets from control subjects. The magnitude of the SRC coefficients against time and frequency are 90% threshold comparisons.

[0198] Figure 12 shows a comparison of frequency analysis of retinal signal data in patients with major depressive disorder (bright spots) and individuals not diagnosed with a mental health condition (dark spots), using a discrete approximation of the Moret waveform as the mother wavelet. The magnitude of the SRC coefficient against time and frequency is a 90% threshold comparison.

[0199] Figure 13 shows a comparison of frequency analysis of retinal signal data in patients with major depressive disorder (bright spots) and individuals not diagnosed with a mental health condition (dark spots), using a discrete approximation of a mother wavelet specifically designed with retinal signal data sets from control subjects. The magnitude of the SRC coefficients against time and frequency is a 90% threshold comparison.

[0200] Figure 14 shows a comparison of frequency analysis of retinal signal data in patients at risk of psychosis (bright spots) and individuals not diagnosed with mental health conditions (dark spots), using a discrete approximation of the Moret waveform as the mother wavelet. The magnitude of the SRC coefficient against time and frequency is a 90% threshold comparison.

[0201] Figure 15 shows a comparison of frequency analysis of retinal signal data in patients at risk of psychosis (bright spots) and individuals not diagnosed with mental health conditions (dark spots), using a discrete approximation of a mother wavelet specifically designed with retinal signal data sets from control subjects. The magnitude of the SRC coefficients against time and frequency is a 90% threshold comparison.

[0202] Figure 16 shows a comparison of frequency analysis of retinal signal data in patients with schizophrenia (bright spots) and patients with bipolar disorder (dark spots), using a discrete approximation of the Moret waveform as the mother wavelet. The magnitude of the SRC coefficient against time and frequency is a 90% threshold comparison.

[0203] Figure 17 shows a comparison of frequency analysis of retinal signal data in patients with schizophrenia (bright spots) and patients with bipolar disorder (dark spots) using a discrete approximation of a mother wavelet specially designed with retinal signal data sets from control subjects. The magnitude of the SRC coefficients against time and frequency are 90% threshold comparisons.

[0204] Figure 18 shows a comparison of frequency analysis of retinal signal data in patients with major depressive disorder (bright spots) and patients with bipolar disorder (dark spots), using a discrete approximation of the Moret waveform as the mother wavelet. The magnitude of the SRC coefficient against time and frequency is a 90% threshold comparison.

[0205] Figure 19 shows a comparison of frequency analysis of retinal signal data in patients with major depressive disorder (bright spots) and patients with bipolar disorder (dark spots) using a discrete approximation of a mother wavelet specially designed with retinal signal data sets from control subjects. The magnitude of the SRC coefficients with respect to time and frequency is a 90% threshold comparison.

[0206] Figure 20 shows a comparison of frequency analysis of retinal signal data in patients with major depressive disorder (bright spots) and patients with schizophrenia (dark spots), using a discrete approximation of the Moret waveform as the mother wavelet. The magnitude of the SRC coefficient against time and frequency is a 90% threshold comparison.

[0207] Figure 21 shows a comparison of frequency analysis of retinal signal data in patients with major depressive disorder (bright spots) and patients with schizophrenia (dark spots) using a discrete approximation of a mother wavelet specially designed with retinal signal data sets from control subjects. The magnitude of the SRC coefficients against time and frequency is a 90% threshold comparison.

[0208] Figure 22 shows a comparison of frequency analysis of retinal signal data in patients with bipolar disorder (bright spots) and patients at risk of psychosis (dark spots), using a discrete approximation of the Moret waveform as the mother wavelet. The magnitude of the SRC coefficient against time and frequency is a 90% threshold comparison.

[0209] Figure 23 shows a comparison of frequency analysis of retinal signal data in patients with bipolar disorder (bright spots) and patients at risk of psychosis (dark spots) using a discrete approximation of a mother wavelet specially designed with retinal signal data sets from control subjects. The magnitude of the SRC coefficients with respect to time and frequency is a 90% threshold comparison.

[0210] Figure 24 shows a comparison of frequency analysis of retinal signal data in patients with schizophrenia (bright spots) and patients at risk of psychosis (dark spots), using a discrete approximation of the Moret waveform as the mother wavelet. The magnitude of the SRC coefficient against time and frequency is a 90% threshold comparison.

[0211] Figure 25 shows a comparison of frequency analysis of retinal signal data in patients with schizophrenia (bright spots) and patients at risk of psychosis (dark spots) using a discrete approximation of a mother wavelet specially designed with retinal signal data sets from control subjects. The magnitude of the SRC coefficients with respect to time and frequency is a 90% threshold comparison.

[0212] Figure 26 shows a comparison of frequency analysis of retinal signal data in patients with major depressive disorder (bright spots) and patients at risk of psychosis (dark spots), using a discrete approximation of the Moret waveform as the mother wavelet. The magnitude of the SRC coefficient against time and frequency is a 90% threshold comparison.

[0213] Figure 27 shows a comparison of frequency analysis of retinal signal data in patients with major depressive disorder (bright spots) and patients at risk of psychosis (dark spots) using a discrete approximation of a mother wavelet specially designed with retinal signal data sets from control subjects. The magnitude of the SRC coefficients with respect to time and frequency is a 90% threshold comparison.

[0214] Figure 28 shows the most distinguishable predictors from feature selection (frequency analysis components in this example) performed by sparse representation-based classification (SRC) for wavelet coefficients at a threshold of p ≤ 0.05 in patients with bipolar disorder and schizophrenia, using a discrete approximation of a mother wavelet specially designed with a retinal signal dataset from control subjects. The colors represent the scale of the magnitude of the SRC at a particular location. Dark spots indicate locations where the SRC is higher than the defined threshold (in the example shown, the threshold is from 70% to 100%).

[0215] Figure 29 shows the most distinguishable predictors from feature selection (frequency analysis components in this example) performed in patients with bipolar disorder and schizophrenia using a mother wavelet specially designed with a retinal signal dataset from control subjects, based on statistical significance at a statistical p-value ≤ 0.05 threshold.

[0216] Figure 30 shows the mapping of the most distinguishable predictors and their significance levels (statistical p-values) from feature selection (frequency analysis components in this example) performed on a principal component analysis classification ranked by statistical significance in patients with bipolar disorder and schizophrenia, using a mother wavelet specially designed with a retinal signal dataset from a control group. In this example, two most distinguishable components were considered at a threshold of p ≤ 0.05. The color (grayscale) is a scale of distinguishing power.

[0217] Figure 31 shows the most distinguishable predictors from feature selection (frequency analysis components in this example) performed by sparse representation-based classification (SRC) of wavelet coefficients at a threshold of p ≤ 0.05 in patients with bipolar disorder and major depressive disorder, using a discrete approximation of a mother wavelet specially designed with a retinal signal dataset from control subjects. Shading indicates the scale of the SRC magnitude at a particular location. Dark spots indicate locations where the SRC is higher than the defined threshold (in the example shown, the threshold is from 70% to 100%).

[0218] Figure 32 shows the most distinguishable predictors from feature selection (frequency analysis components in the illustrated example) performed in patients with bipolar disorder and major depressive disorder using a mother wavelet specially designed with a retinal signal dataset from control subjects, based on statistical significance at a statistical p-value ≤ 0.05 threshold.

[0219] Figure 33 shows the mapping of the most distinguishable predictors and their significance levels (statistical p-values) from feature selection (frequency analysis components in this example) performed on a principal component analysis classification ranked by statistical significance in patients with bipolar disorder and major depressive disorder, using a mother wavelet specially designed with a retinal signal dataset from a control group. In the illustrated example, two most distinguishable components were considered at a threshold of p ≤ 0.05. Shading in grayscale indicates the scale of distinguishing power.

[0220] Figure 34 shows the most distinguishable predictors from feature selection (frequency analysis components in this example) performed by sparse representation-based classification (SRC) of wavelet coefficients at a threshold of p ≤ 0.05 in patients with bipolar disorder and patients at risk of psychosis, using a discrete approximation of a mother wavelet specially designed with a retinal signal dataset from control subjects. Shading indicates the scale of the SRC magnitude at a particular location. Dark spots indicate locations where the SRC is higher than the defined threshold (in the example shown, the threshold is from 70% to 100%).

[0221] Figure 35 shows the most distinguishable predictors from feature selection (frequency analysis components in this example) performed in patients with bipolar disorder and patients at risk of psychosis, based on statistical significance at a statistical p-value ≤ 0.05 threshold, using a specially designed mother wavelet with a retinal signal dataset from control subjects.

[0222] Figure 36 shows the mapping of the most distinguishable predictors and their significance levels (statistical p-values) from feature selection (frequency analysis components in this example) performed on a principal component analysis classification ranked by statistical significance in patients with bipolar disorder and patients at risk of psychosis, using a mother wavelet specially designed with a retinal signal dataset from a control group. In the illustrated example, two most distinguishable components were considered at a threshold of p ≤ 0.05. Shading indicates a scale of distinguishing power.

[0223] Figure 37 shows the most distinguishable predictors from feature selection (frequency analysis components in this example) performed by sparse representation-based classification (SRC) of wavelet coefficients at a threshold of p ≤ 0.05 in patients with schizophrenia and patients at risk of psychosis, using a discrete approximation of a mother wavelet specially designed with a retinal signal dataset from control subjects. Shading indicates the scale of the SRC magnitude at a particular location. Dark spots indicate locations where the SRC is higher than the defined threshold (in the example shown, the threshold is from 70% to 100%).

[0224] Figure 38 shows the most distinguishable predictors from feature selection (frequency analysis components in this example) performed in patients with schizophrenia and patients at risk of psychosis, based on statistical significance at a statistical p-value ≤ 0.05 threshold, using a specially designed mother wavelet with a retinal signal dataset from control subjects.

[0225] Figure 39 shows the mapping of the most distinguishable predictors and their significance levels (statistical p-values) from feature selection (frequency analysis components in this example) performed on a principal component analysis classification ranked by statistical significance in patients with schizophrenia and patients at risk of psychosis, using a mother wavelet specially designed with a retinal signal dataset from a control group. In this example, two most distinguishable components were considered at a threshold of p ≤ 0.05. Shading indicates a scale of distinguishing power.

[0226] Figure 40 shows the most distinguishable predictors from feature selection (frequency analysis components in this example) performed by sparse representation-based classification (SRC) of wavelet coefficients at a threshold of p ≤ 0.05 in patients with major depressive disorder and patients at risk of psychosis, using a discrete approximation of a mother wavelet specially designed with a retinal signal dataset from control subjects. Shading indicates the scale of the SRC magnitude at a particular location. Dark spots indicate locations where the SRC is higher than the defined threshold (in the example shown, the threshold is from 70% to 100%).

[0227] Figure 41 shows the most distinctive predictors from feature selection (frequency analysis components in the illustrated example) performed in patients with major depressive disorder and patients at risk of psychosis, based on statistical significance at a statistical p-value ≤ 0.05 threshold, using a mother wavelet specially designed with a retinal signal dataset from control subjects.

[0228] Figure 42 shows the mapping of the most distinguishable predictors and their significance levels (statistical p-values) from feature selection (frequency analysis components in this example) performed on a principal component analysis classification ranked by statistical significance in patients with major depressive disorder and patients at risk of psychosis, using a mother wavelet specially designed with a retinal signal dataset from a control group. In this example, two most distinguishable components were considered at a threshold of p ≤ 0.05. Shading indicates a scale of distinguishing power.

[0229] Figure 43 shows the mapping of the most distinguishable selected descriptors (spectral entropy in this example) performed by sparse representation-based classification (SRC) at a threshold of p ≤ 0.05 in patients at risk of psychosis and individuals not diagnosed with a mental health condition (control subjects). Dark spots indicate locations of higher distinguishing power at SRCs greater than the defined threshold (in the example shown, the threshold is 80%).

[0230] Figure 44 shows the mapping of the most distinguishable selected descriptors (kurtosis in the illustrated example) performed by sparse representation-based classification (SRC) at a threshold of p ≤ 0.05 in patients at risk of psychosis and individuals not diagnosed with a mental health condition (control group). Dark dots indicate locations of higher distinguishing power at SRCs greater than the defined threshold (the threshold is 80% in the illustrated example).

[0231] Figure 45 shows the mapping of the most distinguishable selected descriptors (Hu moments in this example) performed by sparse representation-based classification (SRC) at a threshold of p ≤ 0.05 in patients at risk of psychosis and individuals not diagnosed with a mental health condition (control subjects). Dark dots indicate locations of higher distinguishing power at SRCs greater than the defined threshold (in the example shown, the threshold is 80%).

[0232] Figure 46 shows the mapping of the most distinguishable selected descriptors (three descriptors in this example: spectral entropy, kurtosis, and Hu moment) performed by sparse representation-based classification (SRC) at a threshold of p ≤ 0.05 in patients at risk of psychosis and individuals not diagnosed with a mental health condition (control subjects). Dark spots indicate locations of higher distinguishivity at SRCs greater than the defined threshold (in the illustrated example, the threshold is 80%). The advantages of additional extractable specific retinal signal descriptors (in both information location and statistical significance) can be seen by comparing the descriptors with those shown in Figures 43, 44, and 45.

[0233] Figure 47 shows the mapping of the most distinguishable selected descriptors (spectral entropy in this example) performed by sparse representation-based classification (SRC) at a threshold of p ≤ 0.05 in patients at risk of psychosis and patients with bipolar disorder. Dark spots indicate locations of higher distinguishing power at SRCs greater than the defined threshold (in the example shown, the threshold is 80%).

[0234] Figure 48 shows the mapping of the most distinguishable selected descriptors (kurtosis in this example) performed by sparse representation-based classification (SRC) at a threshold of p ≤ 0.05 in patients at risk of psychosis and patients with bipolar disorder. Dark dots indicate locations of higher distinguishing power at SRCs greater than the defined threshold (in the example shown, the threshold is 80%).

[0235] Figure 49 shows the mapping of the most distinguishable selected descriptors (Hu moments in this example) performed by sparse representation-based classification (SRC) at a threshold of p ≤ 0.05 in patients at risk of psychosis and patients with bipolar disorder. Dark dots indicate locations of higher distinguishing power at SRCs greater than the defined threshold (in the example shown, the threshold is 80%).

[0236] Figure 50 shows the mapping of the most distinguishable selected descriptors (three descriptors in this example: spectral entropy, kurtosis, and Hu moment) performed by sparse representation-based classification (SRC) at a threshold of p ≤ 0.05 in patients at risk of psychosis and patients with bipolar disorder. Dark spots indicate locations of higher distinguishivity at SRCs greater than the defined threshold (in the illustrated example, the threshold is 80%). The advantages of additional extractable specific retinal signal descriptors (in both information location and statistical significance) can be seen by comparing the descriptors with those shown in Figures 47, 48, and 49.

[0237] Figure 51 shows the mapping of the most distinguishable selected descriptors (spectral entropy in this example) performed by sparse representation-based classification (SRC) at a threshold of p ≤ 0.05 in patients at risk of psychosis and patients with schizophrenia. Dark spots indicate locations of higher distinguishing power at SRCs greater than the defined threshold (in the example shown, the threshold is 80%).

[0238] Figure 52 shows the mapping of the most distinguishable selected descriptors (kurtosis in this example) performed by sparse representation-based classification (SRC) at a threshold of p ≤ 0.05 in patients at risk of psychosis and patients with schizophrenia. Dark dots indicate locations of higher distinguishing power at SRCs greater than the defined threshold (in the example shown, the threshold is 80%).

[0239] Figure 53 shows the mapping of the most distinguishable selected descriptors (Hu moments in this example) performed by sparse representation-based classification (SRC) at a threshold of p ≤ 0.05 in patients at risk of psychosis and patients with schizophrenia. Dark dots indicate locations of higher distinguishing power at SRCs greater than the defined threshold (in the example shown, the threshold is 80%).

[0240] Figure 54 shows the mapping of the most distinguishable selected descriptors (three descriptors in this example: spectral entropy, kurtosis, and Hu moment) performed by sparse representation-based classification (SRC) at a threshold of p ≤ 0.05 in patients at risk of psychosis and patients with schizophrenia. Dark spots indicate locations of higher distinguishivity at SRCs greater than the defined threshold (in the illustrated example, the threshold is 80%). The advantages of additional extractable specific retinal signal descriptors (in both information location and statistical significance) can be seen by comparing the descriptors with those shown in Figures 51, 52, and 53.

[0241] Figure 55 shows the mapping of the most distinguishable selected descriptors (spectral entropy in this example) performed by sparse representation-based classification (SRC) at a threshold of p ≤ 0.05 in patients at risk of psychosis and patients with major depressive disorder. Dark spots indicate locations of higher distinguishing power at SRCs greater than the defined threshold (in the example shown, the threshold is 80%).

[0242] Figure 56 shows the mapping of the most distinguishable selected descriptors (kurtosis in this example) performed by sparse representation-based classification (SRC) at a threshold of p ≤ 0.05 in patients at risk of psychosis and patients with major depressive disorder. Dark spots indicate locations of higher distinguishing power at SRCs greater than the defined threshold (in the example shown, the threshold is 80%).

[0243] Figure 57 shows the mapping of the most distinguishable selected descriptors (Hu moments in this example) performed by sparse representation-based classification (SRC) at a threshold of p ≤ 0.05 in patients at risk of psychosis and patients with major depressive disorder. Dark dots indicate locations of higher distinguishing power at SRCs greater than the defined threshold (in the example shown, the threshold is 80%).

[0244] Figure 58 shows the mapping of the most distinguishable selected descriptors (three descriptors in this example: spectral entropy, kurtosis, and Hu moment) performed by sparse representation-based classification (SRC) at a threshold of p ≤ 0.05 in patients at risk of psychosis and patients with major depressive disorder. Dark spots indicate locations of higher distinguishivity at SRCs greater than the defined threshold (in the illustrated example, the threshold is 80%). The advantages of additional extractable specific retinal signal descriptors (in both information location and statistical significance) can be seen by comparing the descriptors with those shown in Figures 55, 56, and 57.

[0245] It should be clearly understood that not all technical effects described herein are necessarily enjoyed in each and all embodiments of this technology.

[0246] Modifications and improvements to the implementations of this technology described above will be apparent to those skilled in the art. The foregoing description is intended to be illustrative, not restrictive. Therefore, the scope of this technology is intended to be limited only by the appended claims. [Explanation of symbols]

[0247] 100 Computing environment 110 Processor 111 Graphics processing unit 120 Solid state drive 130 Random access memory, memory 140 Display interface 150 Input / output interface 160 System bus, internal and / or external bus 190 Touch screen, screen 192 Touch input / output controller 194 Touch hardware 200 Retinal signal processing system, system 205 Light stimulator 210 Sensor 215 Data collection system 220 Data analysis system 225 Prediction output system

Claims

1. A method performed by at least one processor of a computer system to predict the probability that a patient is in one or more conditions, A step of receiving retinal signal data in response to light stimulation of the patient's retina, wherein the retinal signal data includes one or more of the impedance components of a receiving circuit continuously recorded while the retinal signal data is being captured, the patient's optical parameters, and the parameters of the light stimulation. The steps include extracting one or more retinal signal features from the retinal signal data, The steps include extracting one or more descriptors from the retinal signal features, A step of applying one or more descriptors to a first mathematical model and a second mathematical model, wherein the first mathematical model corresponds to a first state and the second mathematical model corresponds to a second state, thereby generating a first predicted probability for the first state and a second predicted probability for the second state. The steps include outputting the first predicted probability and the second predicted probability, and Methods that include...

2. The method according to claim 1, further comprising the step of displaying an interface including the first predicted probability and the second predicted probability.

3. The method according to claim 1 or 2, further comprising the step of storing the first predicted probability and the second predicted probability in the memory of the computer system.

4. The method according to any one of claims 1 to 3, further comprising the step of collecting the retinal signal data.

5. The steps include obtaining clinical information auxiliary factors extracted from the clinical information corresponding to the aforementioned patient, The steps include applying the aforementioned clinical information auxiliary factors to the first mathematical model and the second mathematical model. The method according to any one of claims 1 to 4, further comprising:

6. The method according to claim 5, wherein the clinical information cofactors correspond to the patient's age, sex, skin pigmentation, or iris color.

7. The method according to any one of claims 1 to 6, wherein the retinal signal data has a sampling frequency between 4 and 24 kHz.

8. The method according to any one of claims 1 to 7, wherein the retinal signal data is collected during a signal acquisition time of 200 milliseconds to 500 milliseconds.

9. The method according to claim 1, wherein the optical parameters include pupil size.

10. The method according to any one of claims 1 to 9, wherein the first state is a medical state and the second state is a control state.

11. The method according to any one of claims 1 to 10, wherein the first or second condition is schizophrenia, bipolar disorder, major depressive disorder, or psychosis.

12. The method according to any one of claims 1 to 10, wherein the first or second condition is post-traumatic stress disorder, stroke, substance abuse, obsessive-compulsive disorder, Alzheimer's disease, Parkinson's disease, multiple sclerosis, autism, or attention deficit disorder.

13. The method according to any one of claims 1 to 12, further comprising the step of receiving user input indicating a selection of the first state and the second state.

14. The method according to any one of claims 1 to 13, further comprising the step of selecting a drug to be administered to the patient based on the first and second predicted probabilities.

15. A system for predicting the probability that a patient is in one or more conditions, comprising at least one processor and a memory for storing a plurality of executable instructions, wherein when the plurality of executable instructions are executed by the at least one processor, the system: A step of receiving retinal signal data in response to light stimulation of the patient's retina, wherein the retinal signal data includes one or more of the impedance components of a receiving circuit continuously recorded while the retinal signal data is being captured, the patient's optical parameters, and the parameters of the light stimulation. The steps include extracting one or more retinal signal features from the retinal signal data, The steps include extracting one or more descriptors from the retinal signal features, A step of applying one or more descriptors to a first mathematical model and a second mathematical model, wherein the first mathematical model corresponds to a first state and the second mathematical model corresponds to a second state, thereby generating a first predicted probability for the first state and a second predicted probability for the second state. The steps include outputting the first predicted probability and the second predicted probability, and A system that performs a method that includes this.

16. The system further comprises a light stimulator and one or more sensors, and when the executable instructions are executed by the at least one processor, the system The light stimulator is to provide a light stimulation signal to the patient's retina, Collecting electrical signals in response to the light stimulus signal via one or more of the sensors, To generate the retinal signal data based on the aforementioned electrical signal. To execute The system according to claim 15.

17. The system further includes a display, and when the executable instructions are executed by the at least one processor, the system causes the system to output an interface via the display that includes the first predicted probability and the second predicted probability. The system according to claim 15 or 16.

18. The memory further includes storing the first predicted probability and the second predicted probability. The system according to any one of claims 15 to 17.

19. Further includes collecting the aforementioned retinal signal data. The system according to any one of claims 15 to 18.

20. To obtain clinical information auxiliary factors extracted from the clinical information corresponding to the aforementioned patient, Applying the aforementioned clinical information auxiliary factors to the first mathematical model and the second mathematical model The system according to any one of claims 15 to 19, further comprising:

21. The aforementioned clinical information cofactors correspond to the patient's age, sex, skin pigmentation, or iris color. The system according to claim 20.

22. The retinal signal data has a sampling frequency between 4 and 24 kHz. The system according to any one of claims 15 to 21.

23. The retinal signal data is collected during a signal acquisition time of 200 milliseconds to 500 milliseconds. The system according to any one of claims 15 to 22.

24. The optical parameters include pupil size. The system according to claim 15.

25. The first state is a medical state, and the second state is a control state. The system according to any one of claims 15 to 24.

26. The first or second condition described above is schizophrenia, bipolar disorder, major depressive disorder, or psychosis. The system according to any one of claims 15 to 24.

27. The first or second condition is post-traumatic stress disorder, stroke, substance abuse, obsessive-compulsive disorder, Alzheimer's disease, Parkinson's disease, multiple sclerosis, autism, or attention deficit disorder. The system according to any one of claims 15 to 24.

28. The further includes receiving user input indicating the selection of the first state and the second state. The system according to any one of claims 15 to 27.

29. Further includes selecting a drug to administer to the patient based on the first and second predicted probabilities. The system according to any one of claims 15 to 28.

Images