Apparatus and method for identification and treatment of patients responsive to antipsychotic drug therapy

Abstract

To provide an apparatus and method for identifying and treating patients responsive to antipsychotic drug therapy.SOLUTION: In some embodiments, a method of treating a patient with an antipsychotic can include identifying the patient as an antipsychotic responder. The method can further include obtaining an electroencephalogram (EEG) signal from the patient. The method can further comprise measuring one or more EEG metrics, thereby identifying the patient as an antipsychotic responder. If the patient is an antipsychotic drug responder, then the method can further comprise administering an antipsychotic drug.SELECTED DRAWING: None

Description

[Technical Field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This application is based on and claims priority to U.S. Provisional Application No. 62 / 895,081, entitled "EEG Biomarkers," filed September 3, 2019, which is incorporated herein by reference in its entirety.

[0002] Technical Field The present disclosure relates to the field of treatment of psychiatric disorders. [Background technology]

[0003] background The treatment of mental disorders such as schizophrenia spectrum or other psychotic disorders often involves the use of psychiatric drugs.Known antipsychotic drugs typically include dopamine receptor agonists.Efforts to develop glutamate receptor agonists generally fail in late-stage clinical trials.There is still a need in the art for a treatment method that includes identifying patients who are antipsychotic drug responders. Summary of the Invention [Means for solving the problem]

[0004] Abstract In some embodiments, a method of treating a patient with an antipsychotic drug can include identifying the patient as an antipsychotic responder. The method can further include obtaining an electroencephalogram (EEG) signal from the patient. The method can further include measuring one or more EEG metrics, thereby identifying the patient as an antipsychotic responder. If the patient is an antipsychotic responder, the method can then further include administering an antipsychotic drug. In some embodiments, the measuring occurs before treatment.

[0005] In some embodiments, the antipsychotic drug is a glutamate receptor agonist. In some embodiments, the antipsychotic drug is a group II metabotropic glutamate receptor (mGluR2 / 3) agonist. In some embodiments, the mGluR2 / 3 agonist is pomaglumetad or a pharmaceutically acceptable salt thereof. In some embodiments, the mGluR2 / 3 agonist is pomaglumetad methionyl or a pharmaceutically acceptable salt thereof.

[0006] In some embodiments, the one or more EEG metrics comprise one or more electrophysiological behavior at one or more brain locations. In some embodiments, the one or more EEG metrics comprise one or more electrophysiological behavior at one or more brain locations under a stimulus of the subject. In some embodiments, the stimulus is a light stimulus, an electrical stimulus, a magnetic stimulus, a tactile stimulus, or an acoustic stimulus.

[0007] In some embodiments, the electrophysiological behavior under stimulation is [Table 1]

[0008] In some embodiments, the one or more EEG metrics include one or more electrophysiological behaviors at resting state at one or more brain locations, the electrophysiological behaviors at the brain locations being: [Table 2]

[0009] In some embodiments, each clinical treatment outcome from a set of clinical treatment outcomes is classified as responsive and non-responsive based on a threshold or a receiver operating characteristic (ROC) curve.

[0010] In some embodiments, the identifying step is performed by a non-transitory processor-readable medium storing code representing instructions for execution by a processor. The code includes code for causing the processor to receive EEG signals recorded from one or more brain locations of the patient. The code can include code for causing the processor to convert the EEG signals into one or more EEG metrics. The code can include further code for causing the processor to execute a model configured to receive the EEG metrics and identify the patient as an antipsychotic responder.

[0011] In some embodiments, the model is a machine learning model. The code can include further code for causing the processor to train the machine learning model based on a training set that includes a set of EEG metrics and a set of clinical treatment outcomes associated with the set of EEG metrics.

[0012] In some embodiments, each clinical outcome from the set of clinical outcomes is determined based on at least one of the MATRICS™ Consensus Cognitive Battery (MCCB™), the Positive and Negative Syndrome Scale (PANSS) score, and the Clinical Global Impression Severity Scale (CGI-S).

[0013] In some embodiments, the set of clinical treatment outcomes comprises a reduction in at least one positive symptom of the PA NSS. In some embodiments, the set of clinical treatment outcomes comprises a reduction in at least one negative symptom of the PA NSS. In some embodiments, an antipsychotic responder is defined by an increase in working memory performance. In some embodiments, an antipsychotic responder is defined by an increase in attention-alertness. In some embodiments, an antipsychotic responder is defined by an increase in reasoning problem solving.

[0014] In some embodiments, the machine learning model comprises a forward propagation machine learning model, a convolutional neural network (CNN), a graph neural network (GNN), an autoencoder, or a transformer neural network. In some embodiments, the machine learning model comprises a logistic regression model, a naive Bayes classifier, a support vector machine (SVM), a random forest, a decision tree, or an extreme gradient boosting (XGBoost) model.

[0015] In some embodiments, the EEG metric includes a power law exponent. In some embodiments, the EEG signal is acquired in the delta band, theta band, alpha band, beta band, or gamma band. In some embodiments, the identifying step identifies the patient as an antipsychotic responder using at most one, at most two, or at most three EEG metrics. In some embodiments, the patient suffers from or is at risk for a psychotic disorder. [Brief explanation of the drawings]

[0016] [Figure 1] FIG. 1 is a schematic illustration of a treatment-response prediction device according to one embodiment.

[0017] [Figure 2] FIG. 2 is a flow chart illustrating a method for treatment-response prediction according to one embodiment.

[0018] [Figure 3] FIG. 3 is a flow chart illustrating a method for treatment-response prediction according to one embodiment.

[0019] [Figure 4] Figure 4 shows the montage for the EEG recording.

[0020] [Figure 5]Figure 5 shows determining the power law exponent (PLE) of the EEG signal.

[0021] [Figure 6] Figure 6 shows the correlation between pre-treatment low-gamma (30 Hz) activity and treatment response based on the MCCB attention-alertness domain score. Patients received photic stimulation at 30 Hz. Color indicates the correlation coefficient r, and the scale is shown on the right. Selected r and p-values ​​are shown in Table 1A.

[0022] [Figure 7] Figure 7 shows the correlation between pre-treatment low beta (15 Hz) activity and treatment response based on the MCCB reasoning problem-solving domain score. Patients received light stimulation at 15 Hz. Color indicates the correlation coefficient r, and the scale is shown on the right. Selected r and p-values ​​are shown in Table 1A.

[0023] [Figure 8] Figure 8 shows the correlation between the pretreatment power law index and treatment response based on the MCCB working memory domain score. EEG readings were taken in the resting state. Color indicates the correlation coefficient r, and the scale is shown on the right. Selected r values ​​and significance are shown in Table 1B.

[0024] [Figure 9] Figure 9 shows the receiver operating curve (ROC) for the effect shown in Figure 8 on EEG lead C3: sensitivity = 0.750, specificity = 0.897 (area under the curve [AUC] = 0.809, p = 0.039). DETAILED DESCRIPTION OF THE INVENTION

[0025] Detailed Description Non-limiting examples of various aspects and variations of these embodiments are described herein and illustrated in the accompanying drawings.

[0026] In one aspect, the present disclosure provides a method of treating a patient with an antipsychotic drug (e.g., a glutamate receptor agonist). The method includes obtaining, or obtaining an electroencephalogram (EEG) signal from the patient, thereby identifying the patient as an antipsychotic responder. The method includes measuring, or measuring one or more EEG metrics, thereby identifying the patient as an antipsychotic responder, and then administering an antipsychotic drug if the patient is an antipsychotic responder.

[0027] In another aspect, one or more embodiments described herein generally relate to devices, methods, and systems for dynamically processing structured and semi-structured data, and in particular to devices, methods, and systems that use models (e.g., neural network models) to efficiently and reliably predict outcomes based on structured and semi-structured data. Treatment-response prediction devices, methods, and systems are disclosed. In some embodiments, treatment-response can be used to process EEG signals in the form of, for example, time series, stationary data, non-stationary data, linear data, non-linear data, etc.

[0028] Described herein are treatment-response prediction devices and methods that predict treatment response based on EEG signals collected from a patient. By enabling patients to be identified as antipsychotic responders prior to treatment, the methods described herein can avoid unnecessary adverse events and side effects of treatment. Furthermore, the methods described herein can increase the response rate to antipsychotics. In certain embodiments, the methods described herein enable the safe and effective use of pomaglumetad, pomaglumetad methionyl, or pharmaceutically acceptable salts thereof in the treatment of psychiatric disorders (e.g., psychotic disorders). In some embodiments, individual EEG metrics that predict antipsychotic (e.g., glutamate receptor agonist) responder status are disclosed. In some embodiments, responder prediction is improved by training a machine learning model on multiple EEG metrics.

[0029] FIG. 1 is a schematic illustration of a treatment-response prediction device 110 according to one embodiment. The treatment-response prediction device 110 can identify a patient as an antipsychotic responder prior to treatment. The treatment-response prediction device 110 can also be configured to execute a model (e.g., an artificial intelligence model) that predicts treatment response based on EEG signals collected for the patient. The set of EEG signals is analyzed by the treatment-response prediction device 110 to generate EEG metrics. The treatment-response prediction device 110 can be coupled to a server computing device 160, a clinician programmer device 170, and / or a patient computing device 180 via a network 150, as needed. The treatment-response prediction device 110, the clinician programmer device 170, and / or the patient computing device 180 can each be a hardware-based computing device and / or a multimedia device, such as, for example, a computer, a desktop, a laptop, a smartphone, a tablet, a wearable device, etc.

[0030] The treatment-response prediction device 110 includes a memory 111, a communication interface 112, and a processor 113. The treatment-response prediction device 110 can receive data including EEG signals from an EEG machine (not shown) that records the patient's brain activity. In some examples, the patient's brain activity can be / include electrical activity, which can be recorded as EEG signals by a set of electrodes connected to the EEG machine, which can be operably coupled to the treatment-response prediction device 110. The EEG machine can transmit a set of EEG signals to the treatment-response prediction device 110. The EEG signals can be recorded in the memory 111 and analyzed by the processor 113 to treat the patient with an antipsychotic medication.

[0031] Network 150 may be a digital telecommunications network of servers and / or computing devices. The servers and / or computing devices on the network may be connected via one or more wired or wireless communication networks (not shown) to share resources, such as data storage and / or computing power. The wired or wireless communication networks between the servers and / or computing devices of network 150 may include one or more communication channels, such as radio frequency (RF) communication channels, extremely low frequency (ELF) communication channels, extremely low frequency (ULF) communication channels, low frequency (LF) communication channels, medium frequency (MF) communication channels, ultra high frequency (UHF) communication channels, extremely high frequency (EHF) communication channels, fiber optic communication channels, electronic communication channels, satellite communication channels, etc. Network 150 may be, for example, the Internet, an intranet, a local area network (LAN), a wide area network (WAN), a metropolitan area network (MAN), a Worldwide Interoperability for Microwave Access Network (WiMAX®), a virtual network, any other suitable communication system, and / or a combination of such networks.

[0032] The server computing device 160 may be / include a computing device medium dedicated to data storage and / or computation purposes, which may include, for example, a network of electronic memory, a network of magnetic memory, a server, a blade server, a storage area network, a network-attached storage device, a deep learning computation server, a deep learning storage server, etc. Each server device 160 may include memory (not shown), a communications interface (not shown), and / or a processor (not shown). The communications interface may receive / transmit data from / to the prediction device 110 via the network 150, the memory may store the data, and the processor may analyze the data. In some examples, the server computing device 160 may be a biobank server that stores data for a long period of time (e.g., 2 years, 5 years, 10 years, 100 years, etc.).

[0033] The clinician computing device 170 and / or the patient computing device 180 can be / include a computing device operably coupled and configured to send and / or receive data and / or analytical models to the treatment-response prediction device 110. A user of the patient computing device 180 and / or the clinician computing device 170 can use the treatment-response prediction device 110 (partially or completely) to select a treatment and / or treatment-response prediction. In some examples, the patient computing device 180 and / or the clinician computing device 170 can be / include a personal computer, laptop, smartphone, custom personal assistant device, etc., each including, for example, memory (not shown), a communications interface (not shown), and / or a processor (not shown). The processor of the patient computing device 180 and / or the clinician computing device 170 can include a hardware-based integrated circuit (IC) or any other suitable processing device configured to implement and / or execute a set of instructions or code. The memory of the patient computing device 180 and / or clinician computing device 170 may include hardware-based charge storage electronic devices or any other suitable data storage medium configured to store data for long-term or batch processing of the data by a processor. The communication interfaces of the patient computing device 180 and / or clinician computing device 170 may include hardware-based devices configured to receive / transmit electrical, electromagnetic, and / or optical signals.

[0034] The memory 111 of the treatment-response prediction device 110 can be, for example, a memory buffer, random access memory (RAM), read-only memory (ROM), hard drive, flash drive, secure digital (SD) memory card, compact disc (CD), external hard drive, erasable programmable read-only memory (EPROM), embedded multi-time programmable (MTP) memory, embedded multimedia card (eMMC), universal flash storage (UFS) device, etc. The memory 111 can store, for example, one or more software modules and / or code including instructions for causing the processor 113 to perform one or more processes or functions (e.g., signal analyzer 114, data preprocessor 115, predictor model 116, etc.).

[0035] The memory 111 may store a set of files associated with the signal analyzer 114, the data preprocessor 115, and / or the predictor model 116 (e.g., generated by executing the signal analyzer 114, the data preprocessor 115, and / or the predictor model 116). The set of files associated with the signal analyzer 114, the data preprocessor 115, and / or the predictor model 116 may include data generated by the signal analyzer 114, the data preprocessor 115, and / or the predictor model 116 during operation of the treatment-response prediction device 110. In some examples, the predictor model 116 may be / include a machine learning model. The machine learning model may store in the memory 111 temporary variables, return memory addresses, variables, the machine learning model's graph (e.g., a set of arithmetic operations or a representation of the set of arithmetic operations used by the machine learning model), graph metadata, assets (e.g., external files), electronic signatures (e.g., specifying the type of machine learning model being exported, and input / output arrays and / or tensors), etc.

[0036] The communication interface 112 of the treatment-response prediction device 110 may include software and / or hardware components (e.g., executed by the processor 113) of the treatment-response prediction device 110 to facilitate data communication between the treatment-response prediction device 110 and external devices (e.g., the server computing device 160, the clinician platform 170, the patient computing device 180, etc.) or internal components (e.g., the memory 111, the processor 113) of the treatment-response prediction device 110. The communication interface 112 is operatively coupled to and used by the processor 113 and / or the memory 111. The communication interface 112 may be, for example, a network interface card (NIC), a Wi-Fi™ module, a Bluetooth® module, an optical communication module, and / or any other suitable wired and / or wireless communication interface. In some examples, the communication interface 112 may facilitate receiving or transmitting data over the network 150. More specifically, in some embodiments, communications interface 112 can facilitate receiving or transmitting data containing EEG signals, models, etc. from / to server computing device 160, clinician platform 170, patient computing device 180, etc. over network 150, each of which is communicatively coupled to treatment-response prediction 110 via network 150. In some examples, communications interface 112 can facilitate receiving or transmitting data from an EEG machine.

[0037] The processor 113 may be a hardware-based integrated circuit (IC) or any other suitable processing device configured to perform or execute a set of instructions or code. For example, the processor 113 may include a general-purpose processor, a central processing unit (CPU), an accelerated processing unit (APU), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), a programmable logic array (PLA), a complex programmable logic device (CPLD), a programmable logic controller (PLC), a graphics processing unit (GPU), a neural network processor (NNP), etc. The processor 113 may be operatively coupled to the memory 111 via a system bus (e.g., an address bus, a data bus, and / or a control bus, not shown).

[0038] The processor 113 includes a signal analyzer 114, a data pre-processor 115, and a predictor model 116. Each of the signal analyzer 114, the data pre-processor 115, and / or the predictor model 116 may include software stored in the memory 111 and executed by the processor 113. For example, code for causing the signal analyzer 114 to fetch / process high-dimensional and large-scale data may be stored in the memory 111 and executed by the processor 113. Alternatively, each of the signal analyzer 114, the data pre-processor 115, and / or the predictor model 116 may be a hardware-based device. For example, the process for the predictor model 116 to predict clinical outcomes may be implemented on a separate integrated circuit chip (e.g., an ASIC).

[0039] The signal analyzer 114 can receive the EEG signal and perform signal analysis on the EEG signal. In some examples, the signal analyzer can perform Fourier analysis (e.g., used for stationary signals) and / or wavelet analysis (e.g., used for non-stationary signals) to transform the EEG signal from the time domain to the frequency domain. The transformed EEG signal can be further analyzed by the signal analyzer 114 to extract significant frequency components of the EEG signal. Fourier analysis of the EEG signal generates a power spectrum, which includes an indication of which frequencies are present in the EEG signal and the relative intensity (or power) of the frequencies. Known power spectrum analyses of EEG or magnetoencephalogram (MEG) signals have not revealed specific frequency peaks that reliably distinguish schizophrenia patients from controls.

[0040] The signal analyzer 114 can further generate EEG metrics. In some examples, the EEG metrics can include a metric based on the ratio of EEG signal power at a first frequency to the EEG signal power at a second frequency (e.g., the ratio of power at 20 Hz to power at 40 Hz). The first and / or second frequencies can be selected from one frequency band or multiple frequency bands. In some examples, the EEG metric can include a power-law exponent (also called "1 / f noise" or "fractal exponent") of the EEG signal. The power spectrum of an EEG signal (in a resting state or under stimulation), when displayed in the frequency domain (i.e., frequency of oscillation is plotted on the x-axis and power is plotted on the y-axis), can be approximated by a straight line when displayed on a log-log plot (see FIG. 5). Mathematically, the power spectrum can be expressed as log(P) = k-βlog(f), or equivalently, Pαf-β (= 1 / fβ), where P represents power, f represents frequency, -β represents the slope of the fitted line, and k represents a constant. The power law exponent β is constant regardless of the resolution at which it is calculated ("scale-invariant" or "fractal" behavior) (see inset in Figure 5). A steeper slope of the power law exponent may reveal a higher degree of "structure" or "memory" in the underlying brain interactions.

[0041] Oscillatory activity at multiple frequencies and brain locations can be observed in the human brain. EEG signals recorded by an EEG machine can be collected for frequency and brain location. Brain location can be based on EEG electrode maps. Frequencies can include the delta band (1-3 cycles per second [Hz]), theta band (4-7 Hz), alpha band (8-12 Hz), beta band (12-30 Hz), and gamma band (40-80 Hz).

[0042] The data pre-processor 115 receives data (e.g., including the signal analyzed by the signal analyzer 114) and can be used to further prepare the data for processing by the predictor model 116. In some implantations, the data pre-processor 115 can normalize the data, feature extraction, size reduction, etc. In some examples, normalizing the data can include amplitude matching, frequency matching, file format (e.g., txt format, CSV format, etc.) adjustment, data format (e.g., comma-separated, semicolon-separated, etc.) adjustment, etc.

[0043] In some examples, the data pre-processor 115 may be configured to receive the set of signals, convert the format of the set of signals, remove measurement artifacts (e.g., generated due to eye blinking or scalp muscle movement of the patient from whom the set of signals is acquired), and / or filter the set of signals (e.g., to reduce noise in the set of signals (denoising)). The data pre-processor 115 may also be configured to perform independent component analysis (ICA) to decompose the set of signals into functionally and spatially separated signals.

[0044] The predictor model 116 (also referred to as a “model”) may be / include a machine learning model, as described in further detail herein. The predictor model 116 may include a forward propagation machine learning model, a convolutional neural network (CNN), a graph neural network (GNN), an autoencoder, a transformer neural network, a logistic regression model, a naive Bayes classifier, a support vector machine (SVM), a random forest, a decision tree, an extreme gradient boosting (XGBoost) model, etc. The predictor model 116 may be configured to include a set of model parameters, including a set of weights, a set of biases, and / or a set of activation functions, that, once trained, may be executed to generate a prediction of a clinical outcome of the EEG signal (e.g., responder, non-responder, 20% responder, 99% responder, response score, etc.). For example, the predictor model 116 may be configured to predict antipsychotic treatment responders.

[0045] In some implantations, clinical outcomes can be classified as "responders" versus "non-responders." Such classification can be defined in several ways, including: • Based on percentage improvement, averaged across all items in the MCCB cognitive assessment battery. • Based on percentage improvement averaged across all items on the Positive Symptoms Scale. • Based on percentage improvement averaged across all items on the negative symptom rating scale. • Based on the average of (i), (ii) and (iii) above and the percentage change in Clinical Global Impression-Severity Scale (CGI-S). In all of the above cases, the percentage improvement from baseline is used as the outcome measure. "Response" can be defined using several different cutoffs (e.g., 20%, 30%, 40%, 50%, and 60% improvement).

[0046] In one example, the predictor model 116 can be / can include a feedforward neural network or deep learning model including an input layer, an output layer, and multiple hidden layers (e.g., 5 layers, 10 layers, 20 layers, 50 layers, 100 layers, 200 layers, etc.), where the multiple hidden layers can include normalization layers, fully connected layers, activation layers, convolutional layers, recurrent layers, and / or any other layers suitable for representing the correlation between EEG signals and clinical outcomes, and each score represents an energy term.

[0047] In one example, the predictor model 116 may be an XGBoost model that includes a set of hyperparameters, such as a number of boosting rounds that defines the number of boosting rounds or trees in the XGBoost model, a max depth that defines the maximum number of nodes allowed from the root of a tree to a leaf of the tree in the XGBoost model, etc. The XGBoost model may include a set of trees, a set of nodes, a set of weights, a set of biases, and other parameters useful for describing the XGBoost model.

[0048] In some embodiments, the predictor model 116 can be configured to iteratively receive EEG signals and / or EEG metrics and generate an output that predicts a clinical outcome (e.g., a binary response where 1 represents a responder and 0 represents a non-responder). The EEG signals and / or EEG metrics can be associated with one clinical outcome. The true clinical outcome can be compared to the output from the predictor model 116 using an optimization model and an objective function (also referred to as a "cost function") to generate a training loss value. The objective function can include, for example, mean squared error, mean absolute error, mean absolute percentage error, log-hyperbolic cosine (logcosh), multi-class cross entropy, etc. A set of model parameters of the predictor model 116 can be modified in multiple iterations, and a first objective function can be run in each iteration until the training loss value converges to a first predetermined training threshold (e.g., 80%, 85%, 90%, 97%, etc.).

[0049] In some embodiments, the predictor model 116 can integrate the EEG metrics and / or EEG signals to generate a composite score that identifies a patient as an antipsychotic responder. In some examples, the composite score can be a normalized range of 0 to 100. A threshold within the normalized range can be set to determine whether a subset of the patient's EEG metrics and / or EEG signals can identify the patient as an antipsychotic responder.

[0050] FIG. 2 is a flowchart illustrating a method 200 of treatment-response prediction according to one embodiment. Method 200 can be performed by a treatment-response prediction device (such as the treatment-response prediction device shown and described in connection with FIG. 1). Method 200 can include receiving 201 electroencephalogram (EEG) signals recorded from one or more brain locations of a patient. Method 200 can further include converting 202 the EEG signals into a set of EEG metrics. The EEG metrics can include electrophysiological behavior at the set of brain locations under stimulation or at rest. The stimuli can include optical, electrical, magnetic, tactile, and / or acoustic stimuli. Method 200 can further include receiving 203 the set of EEG metrics and executing a model to identify the patient as an antipsychotic responder based on the set of EEG metrics. In some embodiments, the model is a machine learning model.

[0051] FIG. 3 is a flowchart illustrating a method 300 of treatment-response prediction according to one embodiment. Method 300 can be performed by a treatment-response prediction device (such as the treatment-response prediction device shown and described in connection with FIG. 1). Method 300 can include receiving 301 electroencephalogram (EEG) signals (e.g., for a set of electrodes). In some examples, the EEG signals can be measured before treatment. In some examples, the EEG signals can be analyzed to calculate power in one or more frequency ranges (also called power bands) of the EEG signals (e.g., delta band, 1-4 Hz; gamma band (30-80 Hz)) and / or fractal exponents (power law exponents). The method 300 may further include determining 302 a set of EEG metrics, where each EEG metric may be identified / generated based on (a) power in a particular frequency band (e.g., delta band, beta band, gamma band, etc.), (b) a ratio of power in two different frequency bands (e.g., first power in beta band / second power in gamma band), (c) a power law exponent, and / or (d) the power of the EEG signal at a drive frequency (i.e., sensory stimulation frequency). The method 300 may further include measuring 303 a clinical outcome indicator after treatment. In some examples, the clinical outcome indicator may be determined by a clinician. In some examples, the clinical outcome indicator may be determined by a treatment-response prediction device based on a set of medical readings.

[0052] The method 300 can further include establishing 304 a statistically significant correlation between the EEG metrics and the clinical outcome indicator. In some examples, statistical significance can be calculated and / or represented by a p-value, e.g., as shown in Tables 1A and 1B. In some examples, a principal component analysis (PCA) can be performed to arrive at a set of principal components (PCs) that contain 75% of the variance in the independent variables (e.g., pre-treatment EEG metrics). In some examples, a multivariate analysis of covariance (MANCOVA) can be performed using all clinical outcomes as dependent variables and the following independent factors: (i) treatment class, indicating treatment with pomaglumetad vs. placebo; (ii) principal components (PCs); and (iii) interactions between (i) and (ii). This determines whether there was a relationship between any principal component and any clinical response, and more importantly, whether there was a significant interaction indicating a difference between treated and untreated subjects. A significance level of p<0.05 can be used.

[0053] In some examples, analysis of covariance (ANCOVA) can be performed on PCs that exhibited significant interactions to determine whether there was a significant relationship (e.g., p<0.05) between the PCs and any number of clinical outcome measures for treated patients. For clinical outcome measures identified in the above steps, the correlation between the outcome measure and the independent variables for that subanalysis (e.g., specific EEG metrics at specific EEG electrodes) can be determined. All independent variables that exhibit a correlation with the outcome measure at a significance level of p<0.05 can be considered for analysis. These can be ranked from lowest (most significant) to highest, and the relationship that exhibits the highest level of significance for each subanalysis can be examined in more detail by constructing a topographic cortical map that includes electrode-level correlations and significance levels for that independent variable-dependent variable (i.e., EEG metric-clinical outcome measure) pair.

[0054] Method 300 may further include selecting 305 a subset of correlations from the statistically significant correlations having high statistical significance and effect size (e.g., large correlation coefficients). Method 300 may further include training 306 a machine learning model based on the subset of correlations. After training, method 300 may further include running 307 the machine learning model to generate a clinical outcome based on the pattern of EEG metrics.

[0055] In one aspect, the present disclosure provides a method for treating a patient with an antipsychotic drug. The method includes obtaining or obtaining an electroencephalogram (EEG) signal from the patient, thereby identifying the patient as an antipsychotic (e.g., glutamate receptor agonist) responder. The method includes measuring or measuring one or more EEG metrics, thereby identifying the patient as an antipsychotic responder, and then administering an antipsychotic drug if the patient is an antipsychotic responder.

[0056] As used herein, the term "patient" refers to a human subject suffering from or at risk of suffering from a mental disorder. A patient exhibits one or more symptoms of a mental disorder. Exemplary mental disorders are those listed in the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) (5), which is incorporated by reference for purposes of defining mental disorders and symptoms by which a patient can be identified as suffering from or at risk of suffering from a mental disorder. th (ed.) (2013). In some embodiments, the psychiatric disorder is schizophrenia spectrum or other psychotic disorder.

[0057] According to the methods of the present disclosure, patients can be identified as potential responders to antipsychotic medications (e.g., glutamate receptor agonists). Other factors, including symptoms of the psychiatric disorder, can be considered by the treating physician or other healthcare professional. It is understood that antipsychotic medications can be used to treat other disorders, and the methods of the present disclosure are not limited to schizophrenia spectrum or other psychotic disorders. Without being bound by theory, it is believed that the EEG metrics disclosed herein predict response through correlations between EEG signals and underlying human brain biochemistry. The characteristics of brain physiology underlying responsiveness to treatment in psychiatric patients may extend to other disorders, including neurodevelopmental disorders, bipolar and related disorders, depressive disorders, anxiety disorders, and the like.

[0058] Given the observed prediction of response to treatment in certain clinical domains (attention-alertness, reasoning problem solving, and working memory), the methods of the present disclosure may be applied to psychiatric disorders affecting these clinical domains, including but not limited to schizophrenia spectrum and other psychotic disorders.

[0059] Exemplary antipsychotic drugs (known in the art as "typical antipsychotics") that can be administered according to the methods of the present disclosure, or to which responders can be identified, include, but are not limited to, chlorpromazine, fluphenazine, haloperidol, loxapine, perphenazine, pimozide, thioridazine, thiothixene, trifluoperazine.

[0060] Additional exemplary antipsychotic agents (known in the art as "atypical antipsychotics") that can be administered according to the methods of the present disclosure, or to which responders can be identified, include, but are not limited to, aripiprazole (commercially available as Abilify), asenapine (commercially available as Saphris), clozapine (commercially available as Clozaril), iloperidone (commercially available as Fanapt), lurasidone (commercially available as Latuda), olanzapine (commercially available as Zyprexa), olanzapine / fluoxetine (commercially available as Symbyax), paliperidone (commercially available as Invega), paliperidone, pimavanser, quetiapine (commercially available as Seroquel), risperidone (commercially available as Risperdal), ziprasidone (commercially available as Geodon), or derivatives thereof.

[0061] Glutamate receptor agonists have previously been unable to achieve the clinical benefits predicted from preclinical studies. The disclosed methods, along with associated computational methods and instruments, provide successful treatment with glutamate receptor agonists that would not otherwise be safe and effective.

[0062] Exemplary glutamate receptor agonists include D-serine, CTP-692 (deuterated D-serine), SAGE-718 (a positive allosteric modulator [PAM] at the NMDA receptor), sarcosine (a GlyT-1 inhibitor; also increases glycine), LY379268, eglumegumad, pomaglumetad (LY2140023), and pomaglumetad methionyl, or pharmaceutically acceptable salts thereof.

[0063] In some embodiments, the glutamate receptor agonist is D-serine or a pharmaceutically acceptable salt thereof.

[0064] In some embodiments, the glutamate receptor agonist is CTP-692 or a pharmaceutically acceptable salt thereof.

[0065] In some embodiments, the glutamate receptor agonist is SAGE-718 or a pharmaceutically acceptable salt thereof.

[0066] In some embodiments, the glutamate receptor agonist is a GlyT-1 inhibitor or a pharmaceutically acceptable salt thereof. In some embodiments, the glutamate receptor agonist is bifoperfin, PF-3463275, GSK1018921, Org25935, AMG747, SSR504734, SSR103800, DCCCyB, R231857, R213129, ASP2535, or any derivative thereof, or a pharmaceutically acceptable salt thereof. The chemical structures of these molecules are provided below. [ka]

[0067] In some embodiments, the glutamate receptor agonist is sarcosine or a pharmaceutically acceptable salt thereof. Sarcosine is a GlyT-1 inhibitor; sarcosine also increases glycine.

[0068] In some embodiments, the glutamate receptor agonist is LY379268 or a pharmaceutically acceptable salt thereof.

[0069] In some embodiments, the glutamate receptor agonist is eglumegad or a pharmaceutically acceptable salt thereof.

[0070] In some embodiments, the glutamate receptor agonist is pomaglumetad or a pharmaceutically acceptable salt thereof.

[0071] In some embodiments, the glutamate receptor agonist is pomaglumetad methionyl or a pharmaceutically acceptable salt thereof.

[0072] Pomaglumetad is an amino acid analog drug that acts as a highly selective agonist for the metabotropic glutamate receptor group II subtypes mGluR2 and mGluR3. Because pomaglumetad exhibits poor oral absorption and bioavailability in humans, human studies investigating the therapeutic use of pomaglumetad have focused on the prodrug, pomaglumetad methionyl. The dose of pomaglumetad methionyl given to patients varied across clinical trials, but doses typically ranged from 10 mg to 40 mg twice daily (BID). In early phase II monotherapy trials, the dose shown to be effective was 40 mg BID. Subsequent studies investigating the use of pomaglumetad methionyl as an adjunct to antipsychotic medications already being used by patients participating in these studies used a lower dose of 20 mg BID. If treatment was well tolerated after 1 week at this target dose, the dose was increased to 40 mg BID. However, if the 20 mg dose was not well tolerated, the dose was reduced to 10 mg.

[0073] Clinical trials of pomaglumetad methionil were discontinued because glutamate receptor agonists were not significantly more effective than placebo, as determined using the PANSS total score. The disclosed devices and methods relate to the surprising discovery that EEG metrics, alone or in combination with the use of machine learning-based models, can select patients who will respond to glutamate receptor agonist (e.g., pomaglumetad methionil) therapy and / or predict clinical outcomes based on EEG metrics.

[0074] Depending on the specific condition being treated, such agents may be formulated into liquid (e.g., solutions, suspensions, or emulsions) or solid dosage forms (capsules or tablets) and may be administered systemically or locally. The agents may be delivered, for example, in timed-, controlled-, or sustained-release forms, as known to those skilled in the art. Techniques for formulation and administration are described in Remington: The Science and Practice of Pharmacy (2007). th (ed.) Lippincott, Williams & Wilkins (2000). Suitable routes may include oral, buccal, by inhalation spray, sublingual, rectal, transdermal, vaginal, transmucosal, nasal or intestinal administration; intramuscular, subcutaneous, intramedullary injection, and parenteral delivery, including intrathecal, direct intraventricular, intravenous, intraarticular, intrasternal, intrasynovial, intrahepatic, intralesional, intracranial, intraperitoneal, intranasal, or intraocular injection, or other modes of delivery. In some embodiments, the pharmaceutical composition is administered orally. In some embodiments, the pharmaceutical composition is administered intravenously. In some embodiments, the pharmaceutical composition is administered intramuscularly. In some embodiments, the pharmaceutical composition is administered intrathecally. In some embodiments, the pharmaceutical composition is administered subcutaneously.

[0075] The pharmaceutical compositions or combinations of the present invention may be in unit dosage form (e.g., tablets, capsules, caplets or particulates), and the appropriate dosage of the active ingredient may vary depending on various factors such as age, weight, sex, route of administration or salt used, for example.

[0076] Generally, the methods of treatment disclosed herein result in a reduction in the severity of a disease or condition in a subject. The term "reduce" means to inhibit, suppress, attenuate, weaken, arrest, or stabilize the symptoms of a disease or condition.

[0077] As used herein, the terms "treat," "treating," "treatment," or "therapy" refer to obtaining a beneficial or desired result, e.g., a clinical result. Beneficial or desired results can include, but are not limited to, alleviating one or more symptoms of schizophrenia as defined herein, e.g., alleviating a positive symptom of schizophrenia or a negative symptom of schizophrenia as defined herein. One aspect of treatment is, for example, that the treatment should have minimal adverse effects on the patient, e.g., a high level of safety. The term "alleviation," as used herein, refers to, for example, reducing at least one of the frequency and magnitude of symptoms of the condition in a patient, e.g., with respect to symptoms of the condition. In one embodiment, the term "method of treatment" as used herein refers to a "method for treating."

[0078] In some embodiments of the disclosed methods, the antipsychotic agent (eg, a glutamate receptor agonist) is administered in an amount effective to produce the desired therapeutic effect (ie, a therapeutically effective amount).

[0079] The term "antipsychotic drug" as used herein refers to a neuroleptic drug used to treat psychotic disorders such as schizophrenia. In one embodiment, the antipsychotic drug is selected from the group including, for example, typical antipsychotic drugs and atypical antipsychotic drugs. In another embodiment, the antipsychotic drug is a typical antipsychotic drug. In yet another embodiment, the antipsychotic drug is an atypical antipsychotic drug.

[0080] The term "typical antipsychotic" as used herein refers to a first-generation antipsychotic selected from the group including, for example, butyrophenones (e.g., haloperidol), diphenylbutylpiperidines (e.g., pimozide), phenothiazines (e.g., chlorpromazine, fluphenazine, perphenazine, prochlorperazine, trifluoperazine), and thioxanthenes (e.g., thiothixene). In one embodiment, the typical antipsychotic is selected from the group including haloperidol, pimozide, chlorpromazine, fluphenazine, perphenazine, prochlorperazine, trifluoperazine, and thiothixene, or a salt thereof.

[0081] The term "atypical antipsychotic" as used herein refers to a second-generation antipsychotic selected from the group including, for example, benzamides (e.g., sultopride), benzisoxazoles / benzisothiazoles (e.g., lurasidone, paliperidone, risperidone), phenylpiperazines / quinolinones (e.g., aripiprazole, brexpiprazole, cariprazine), tricyclic antidepressants (e.g., clozapine, olanzapine, quetiapine, asenapine, zotepine). In one embodiment, the atypical antipsychotic is selected from the group including sultopride, lurasidone, paliperidone, risperidone, brexpiprazole, cariprazine, clozapine, olanzapine, quetiapine, asenapine, and zotepine, or a salt thereof. [Example]

[0082] Example The following specific examples are illustrative, and do not limit the scope of the claimed invention.

[0083] method Pretreatment After receiving the data, the data format can be converted (e.g., file format change). The data can be further processed to remove and filter measurement artifacts (e.g., due to eye blinks or scalp muscle movements). Additionally or alternatively, independent component analysis (ICA) can be performed. ICA is a process that allows recorded EEG data to be decomposed into functionally and spatially separated signals (Onton et al., (2006) Neurosci Biobehav Rev 30(6):808-822), which can also help denoise the signal. The following analysis techniques can be applied to the filtered data, free of artifacts, as well as to the estimated sources revealed by ICA.

[0084] Power Spectral Analysis Oscillatory activity at many different frequencies has been observed in the human brain. These are conventionally divided into the slow frequency ranges of delta (1–3 cycles per second [Hz]) and theta (4–7 Hz); the mid-frequency, alpha (8–12 Hz) and beta (12–30 Hz); and the gamma band (40–80 Hz).

[0085] Fourier analysis of EEG signals produces what is called a power spectrum, i.e., a measure of which frequencies are present and their relative intensities (or power). To our knowledge, this power spectrum analysis has never been used successfully to predict drug response or to subclassify schizophrenic patients.

[0086] In some examples, wavelet analysis can be used to analyze EEG signals. Wavelet analysis provides similar information but uses a rolling time window to determine the frequencies present. This is particularly well suited to shorter (<10 seconds) sets of data segments that may occur after artifact removal.

[0087] Power-law behavior The power spectrum of a resting-state EEG signal, when displayed in the frequency domain (i.e., with frequency of oscillation on the x-axis and power on the y-axis), forms an approximately straight line when displayed on a log-log plot; see Figure 5 (from Miller et al. (2019) PLoS Comput Biol 5(12):e1000609).

[0088] Mathematically, this can be expressed as log(P) = k-βlog(f), or equivalently, Pαf-β (= 1 / fβ), where P = power, f = frequency, -β is the slope of the fitted line, and k is a constant. This phenomenon is known by various terms, such as "power law" spectra or "1 / f noise." Because the power law exponent β is constant regardless of the resolution at which it is calculated, this is often referred to as "scale-invariant" or "fractal" behavior (Lowen and Teich, 2005).

[0089] Treatment response Response to treatment is determined by calculating the change from baseline in key clinical outcome measures as follows:

[0090] Positive and Negative Symptom Scale (PANSS). Correlations consist of the positive subscale (range 7-49), negative subscale (7-49), general psychopathology subscale (16-112), and PANNS total score (30-210).

[0091] Clinical Global Impression-Severity Scale (CGI-S), which measures the overall severity of a patient's symptoms and ranges from 1 to 7.

[0092] The 16-item Negative Symptom Assessment (NSA-16) uses a total score ranging from 16 to 96.

[0093] Measurement and Treatment Studies to Improve Cognition in Schizophrenia: The Unified Cognitive Assessment Battery (MCCB) assesses cognitive function in seven domains: speed of processing, working memory, verbal learning, visual learning, reasoning and problem solving, attention / alertness, and social cognition.

[0094] The Personal and Social Performance (PSP) scale ranges from 1 to 100 and uses a total score assessing four domains of functioning.

[0095] Example 1 Pomaglumetad methionil (LY-2140023, or "POMA") is an experimental antipsychotic drug that is an agonist of metabotropic (mGluR2 / 3) glutamate receptors and has no known effects on dopamine receptors. This profile differs from all currently used antipsychotics, which act on the dopamine (DA) system. Multiple Phase II and Phase III clinical trials have suggested that while this drug may not be effective for all schizophrenic patients, there may be specific subgroups for whom this drug is uniquely useful. Our goal was to develop a novel EEG biomarker to identify schizophrenic patients more likely to respond positively to treatment with POMA. Previous attempts to use EEG readings to predict antipsychotic treatment responders have largely failed. These studies generally investigated DA-agonist antipsychotics. Furthermore, we investigated additional EEG measures, e.g., the magnitude of the power law exponent (PLE), the response to light stimulation, which, to our knowledge, has not been used in previous predictive studies.

[0096] method This study used data from clinical trials NCT00845026 (N = 117) and NCT01052103 (N = 196) to study male and female patients with schizophrenia treated with antipsychotic medication versus standard of care. EEG recordings were taken during the pretreatment period using a standard 19-lead montage (Figure 4) both in the resting state and when patients were exposed to photic stimulation (light flashes) at frequencies ranging from 1 to 30 Hz. We subjected resting EEG data to known power spectral analysis to determine the intensity of activity in the delta (1-4 Hz), theta (4-7 Hz), alpha (8-13 Hz), beta (13-30 Hz), and gamma (30-80 Hz) frequency bands. We calculated only the power at the stimulated frequencies and, for resting-state recordings, the power law index (PLE; also known as the "fractal index"). The power spectrum of a resting-state EEG signal, when displayed in the frequency domain, i.e., graphing power as a function of oscillation frequency, often forms an approximately straight line when displayed on a log-log plot, with the slope of the fitted line being the PLE (see, e.g., Figure 5). We calculated all predictive measures for each EEG electrode.

[0097] Response to POM treatment was reported as a percentage change from baseline to study endpoint on several clinical outcome measures, including the Positive and Negative Symptom Scale (PANSS) and the MATRICS Consensus Cognitive Assessment Battery (MCCB) and its seven individual cognitive domain scales. We performed statistical analyses to determine whether significant relationships existed between any of our computed EEG metrics in the pretreatment condition and treatment outcome. Positive symptoms of the PANNSS include delusions, impaired conceptual integration, hallucinations, agitation, grandiosity, suspiciousness / paranoia, and hostility. Negative symptoms of the PANNSS include flat affect, emotional withdrawal, impaired communication, social withdrawal due to passivity / hypoactivity, difficulty with abstract thinking, lack of spontaneity and fluency in speech, and stereotyped thinking.

[0098] result Surprisingly, we identified several pre-treatment EEG metrics that correlated with clinical outcome and predicted response to treatment with POMA (Table 1A and Table 1B). [Table 1A] [Table 1B]

[0099] For example: (1) In the 30 Hz light-stimulated condition, there was a positive correlation between pre-treatment low-gamma (30 Hz) activity in EEG lead T6 and post-treatment improvement in cognition, as measured by the Attention-Alertness domain score of the MCCB (r = 0.385, p = 0.000001) (Table 1A, line 6). The correlation coefficients ranged from 0.34 to 0.39, with similar levels of significance in other occipital and inferior leads (Figure 6). (2) In the 15 Hz light-stimulated condition, there was a positive correlation between pre-treatment low-beta (15 Hz) activity in EEG lead T5 and post-treatment improvement in cognition, as measured by the Inference-Problem-Solving domain score of the MCCB (r = 0.244, p = 0.004932) (Table 1A, line 10). The correlation coefficients ranged from 0.19 to 0.24, with similar levels of significance in other left temporo-parietal leads (Figure 7). (3) For the left central electrode C3, a positive correlation existed between pretreatment PLE and improvement in the MCCB working memory (WM) domain score (r = 0.288, p = 0.000558) (Table 1B, line 17). Correlation coefficients of 0.25 to 0.31 were found in other fronto-central leads, with similar significance levels (Figure 8).

[0100] Taking this effect (pretreatment PLE at left central electrode C3) as an example, and using a 50% improvement in WM performance as the definition of treatment response, receiver operating curve (ROC) analysis revealed that PLE at lead C3 could identify POM responders with a sensitivity of 0.750 and a specificity of 0.897 (AUC = 0.809, p = 0.039) (Figure 9). In total, 23 lead-level relationships were determined to have relatively high effect sizes and robust statistical significance, and thus potential clinical utility (Tables 1A and 1B). Notably, all response variables we identified included improvement in cognitive function, rather than positive or negative symptoms or other measures of psychopathology. Positive EEG predictors did not emerge within a single specific cortical region.

[0101] In summary, this data demonstrates the existence of patient subgroups that show unique benefits in terms of cognitive improvement and can be characterized by specific EEG "spectral fingerprints."

[0102] Example 2 The inventors have identified individual EEG-based pretreatment metrics that serve as prospective biomarkers of treatment outcome. The methodology described herein can be applied to other psychotropic drugs and various conditions other than schizophrenia, as these drugs may also result in unique patterns of EEG activity in responders versus non-responders that may be difficult to understand using known methods alone. In the treatment setting, there is clear value in identifying those likely to respond to a drug, thereby reducing the time and resources spent on pharmacological trial and error. Furthermore, the types of markers developed by the inventors can be used to narrow patient samples for clinical trials, potentially leading to smaller studies and shorter trials, as well as overall lower drug development costs.

[0103] Example 3 Specific brain phenotypes that respond uniquely to POMA are characterized by a combination of the effects described in Examples 1 and 2. To identify such patterns, we developed an artificial intelligence (AI) approach using deep learning artificial neural networks (ANNs). This methodology is well suited to complex, nonlinear pattern recognition tasks involving multiple inputs. The resulting "composite biomarker" takes into account all of the identified effects and is a more robust predictor than any single predictor.

[0104] We propose to create a deep neural network consisting of four layers: an input layer of 23 nodes, each corresponding to one of the biomarkers in Table 1; two hidden layers, hidden layer 1 of 33 nodes and hidden layer 2 of 7 nodes; and an output layer of one node that represents a "responder" when active and a "non-responder" when inactive. The model can be a four-layer model, where the number of nodes in the first and second hidden layers is

number

[0105] Achieving an optimal network architecture (e.g., number of layers, number of nodes per layer) can be approached as an optimization problem. Several different methodologies have been proposed to address this problem, including the "evolutionary approach," the "constructive approach," or the "pruning approach" (Thomas and Suhner, 2015), which is the method we chose to use. According to this approach, a large network is started. During the training process, it may become clear that certain parameters are not being utilized (e.g., some connection weights are zero or near zero). These are then removed. While this approach can yield very effective networks, the drawback is that the process can be computationally intensive. Given the 72-processor computer cluster that our lab owns and will be used to run these models, this is not a significant consideration. If the pruning approach fails to produce suitable results, other model-building approaches can be used, which is necessarily a trial-and-error process. The present invention provides, for example, the following items. (Item 1) 1. A method of treating a patient with an antipsychotic drug, comprising: obtaining or having obtained electroencephalogram (EEG) signals from said patient; and measuring or having measured one or more EEG metrics, identifying said patient as an antipsychotic responder; identifying the patient as an antipsychotic responder by if the patient is an antipsychotic responder, then administering the antipsychotic; A method comprising: (Item 2) 10. The method of claim 1, wherein the measuring is performed before treatment. (Item 3) 3. The method of claim 1 or 2, wherein the antipsychotic drug is a glutamate receptor agonist. (Item 4) 4. The method of claim 3, wherein the antipsychotic drug is a group II metabotropic glutamate receptor (mGluR2 / 3) agonist. (Item 5) 5. The method of claim 4, wherein the mGluR2 / 3 agonist is pomaglumetad or a pharmaceutically acceptable salt thereof. (Item 6) 5. The method of claim 4, wherein the mGluR2 / 3 agonist is pomaglumetad methionyl or a pharmaceutically acceptable salt thereof. (Item 7) 7. The method of any one of items 1 to 6, wherein the one or more EEG metrics comprise one or more electrophysiological behaviors at one or more brain locations. (Item 8) 8. The method of any one of items 1 to 7, wherein the one or more EEG metrics comprise one or more electrophysiological behaviors at one or more brain locations under stimulation of the subject. (Item 9) 9. The method of claim 8, wherein the stimulus is an optical stimulus, an electrical stimulus, a magnetic stimulus, a tactile stimulus, or an acoustic stimulus. (Item 10) The electrophysiological behavior under stimulation is [Table 3] 10. The method according to item 8 or 9, selected from: (Item 11) The one or more EEG metrics include one or more electrophysiological behaviors at one or more brain locations in a resting state, the electrophysiological behaviors at the brain locations being: [Table 4] The method according to any one of items 1 to 7, wherein the method is selected from the group consisting of: (Item 12) 12. The method of any one of items 1 to 11, wherein each clinical treatment outcome from the plurality of clinical treatment outcomes is classified as responsive and non-responsive based on a threshold or a receiver operating characteristic (ROC) curve. (Item 13) The identifying step is performed by a non-transitory processor-readable medium storing code representing instructions to be executed by a processor, the code causing the processor to: receiving the EEG signals recorded from the one or more brain locations of the patient; converting the EEG signals into the one or more EEG metrics; and 13. The method of any one of items 1 to 12, comprising code for receiving the EEG metrics and running a model configured to identify the patient as an antipsychotic responder. (Item 14) the model is a machine learning model, and the non-transitory processor-readable medium comprises: 14. The method of claim 13, further comprising: code for training the machine learning model based on a training set comprising a plurality of EEG metrics and a plurality of clinical treatment outcomes associated with the plurality of EEG metrics. (Item 15) 14. The method of item 13, wherein each clinical treatment outcome from the plurality of clinical treatment outcomes is determined based on at least one of the MATRICS™ Convergence Cognitive Assessment Battery (MCCB™), Positive and Negative Symptom Scale (PANSS) score, and Clinical Global Impression-Severity Scale (CGI-S). (Item 16) Item 16. The method of item 15, wherein the plurality of clinical treatment outcomes comprises a reduction in at least one positive symptom of the PANNSS. (Item 17) 16. The method of item 15, wherein the plurality of clinical treatment outcomes comprises a reduction in at least one negative symptom of the PANNSS. (Item 18) 2. The method of item 1, wherein the antipsychotic responder is defined by an increase in working memory performance. (Item 19) 2. The method of item 1, wherein the antipsychotic responder is defined by increased attention-alertness. (Item 20) 2. The method of item 1, wherein the antipsychotic responder is defined by improved reasoning problem solving. (Item 21) Item 14. The method of item 13, wherein the machine learning model comprises a feedforward propagation machine learning model, a convolutional neural network (CNN), a graph neural network (GNN), an autoencoder, or a transformer neural network. (Item 22) Item 14. The method of item 13, wherein the machine learning model comprises a logistic regression model, a naive Bayes classifier, a support vector machine (SVM), a random forest, a decision tree, or an extreme gradient boosting (XGBoost) model. (Item 23) 2. The method of claim 1, wherein the EEG metric comprises a power law index. (Item 24) 2. The method of claim 1, wherein the EEG signals are acquired in the delta band, theta band, alpha band, beta band or gamma band. (Item 25) 13. The method of any one of items 1 to 12, wherein the identifying step identifies the patient as an antipsychotic responder using at most one, at most two, or at most three EEG metrics. (Item 26) 26. The method of any one of items 1 to 25, wherein the patient is suffering from or at risk of a psychotic disorder. (Item 27) 1. A non-transitory processor-readable medium storing code representing instructions for execution by a processor, the code causing the processor to: receiving electroencephalogram (EEG) signals recorded from one or more brain locations of the patient; converting the EEG signals into one or more EEG metrics; and a non-transitory processor-readable medium comprising code for executing a model configured to receive the one or more EEG metrics and identify the patient as an antipsychotic responder based on the one or more EEG metrics; (Item 28) 28. The non-transitory processor-readable medium of item 27, wherein the antipsychotic drug is a glutamate receptor agonist. (Item 29) Item 28. The non-transitory processor-readable medium of item 27, wherein the EEG signals are recorded before treatment. (Item 30) 28. The non-transitory processor-readable medium of item 27, wherein the antipsychotic agent is a group II metabotropic glutamate receptor (mGluR2 / 3) agonist. (Item 31) 31. The non-transitory processor-readable medium of item 30, wherein the mGluR2 / 3 agonist is pomaglumetad or a pharmaceutically acceptable salt thereof. (Item 32) 31. The non-transitory processor-readable medium of item 30, wherein the mGluR2 / 3 agonist is pomaglumetad methionyl or a pharmaceutically acceptable salt thereof. (Item 33) 28. The non-transitory processor-readable medium of item 27, wherein the one or more EEG metrics include a power law index. (Item 34) 28. The non-transitory processor-readable medium of item 27, further comprising code for recording EEG signals from the patient. (Item 35) removing measurement artifacts from the EEG signals before the EEG signals are converted, including periods when the patient moves and periods when the patient blinks their eyes; and performing independent component analysis (ICA) before the EEG signal is transformed to decompose the EEG signal and remove noise from the EGG signal; 30. The non-transitory processor-readable medium of item 29, further comprising code for: (Item 36) Item 28. The non-transitory processor-readable medium of item 27, wherein recording the EEG signals is in a resting state. (Item 37) 28. The non-transitory processor-readable medium of item 27, wherein recording the EEG signal is during exposure to a stimulus. (Item 38) Item 38. The non-transitory processor-readable medium of item 37, wherein the stimulus is a light stimulus. (Item 39) Item 38. The non-transitory processor-readable medium of item 37, wherein the stimulus is an electrical stimulus, a magnetic stimulus, a tactile stimulus, or an acoustic stimulus. (Item 40) the model is a machine learning model, and the non-transitory processor-readable medium comprises: 28. The non-transitory processor-readable medium of claim 27, further comprising code for training the machine learning model based on a training set comprising a plurality of EEG metrics and a plurality of clinical treatment outcomes associated with the plurality of EEG metrics, the plurality of EEG metrics comprising the one or more EEG metrics. (Item 41) 41. The non-transitory processor-readable medium of claim 40, wherein each clinical treatment outcome from the plurality of clinical treatment outcomes is classified as responsive and non-responsive based on a threshold or a receiver operating characteristic (ROC) curve. (Item 42) Item 41. The non-transitory processor-readable medium of item 40, wherein each clinical treatment outcome from the plurality of clinical treatment outcomes is classified as responsive and non-responsive based on a receiver operating characteristic (ROC) curve. (Item 43) Item 41. The non-transitory processor-readable medium of Item 40, wherein each clinical treatment outcome from the plurality of clinical treatment outcomes is determined based on at least one of the MATRICS™ Convergence Cognitive Assessment Battery (MCCB™), a scale for rating positive symptoms, a scale for rating negative symptoms, and a Clinical Global Impression-Severity scale (CGI-S). (Item 44) 41. The non-transitory processor-readable medium of claim 40, wherein the plurality of clinical treatment outcomes comprises a reduction in at least one positive symptom of the PANNSS. (Item 45) Item 41. The non-transitory processor-readable medium of Item 40, wherein the plurality of clinical treatment outcomes comprises a reduction in at least one negative symptom of the PANNSS. (Item 46) 28. The non-transitory processor-readable medium of item 27, wherein the antipsychotic responder is defined by an increase in working memory performance. (Item 47) Item 28. The non-transitory processor-readable medium of item 27, wherein the antipsychotic responder is defined by increased attention-alertness. (Item 48) 28. The non-transitory processor-readable medium of item 27, wherein the antipsychotic responder is defined by improved reasoning problem solving. (Item 49) 28. The non-transitory processor-readable medium of item 27, wherein the antipsychotic responder is defined by working memory performance. (Item 50) Item 51. The non-transitory processor-readable medium of item 40, wherein the machine learning model comprises a feedforward propagation machine learning model, a convolutional neural network (CNN), a graph neural network (GNN), an autoencoder, or a transformer neural network. Item 41. The non-transitory processor-readable medium of item 40, wherein the machine learning model comprises a logistic regression model, a naive Bayes classifier, a support vector machine (SVM), a random forest, a decision tree, or an extreme gradient boosting (XGBoost) model. (Item 52) 28. The non-transitory processor-readable medium of item 27, wherein the EEG signals are acquired in the delta band, theta band, alpha band, beta band, or gamma band. (Item 53) 28. The non-transitory processor-readable medium of item 27, wherein the one or more EEG metrics include one or more electrophysiological behaviors at one or more brain locations under stimulation. (Item 54) Item 54. The non-transitory processor-readable medium of Item 53, wherein the stimulus is an optical stimulus, an electrical stimulus, a magnetic stimulus, a tactile stimulus, or an acoustic stimulus. (Item 55) The electrophysiological behavior under stimulation is [Table 5] Item 54. The non-transitory processor-readable medium of item 53, selected from: (Item 56) The one or more EEG metrics include one or more electrophysiological behaviors at one or more brain locations in a resting state, the electrophysiological behaviors at the brain locations being: [Table 6] 28. The non-transitory processor-readable medium of item 27, selected from: (Item 57) 56. The non-transitory processor-readable medium of any one of items 27 to 55, wherein the patient is suffering from or at risk for a psychotic disorder.

Claims

[Claim 1] The invention described in this specification.

Images