Methods and related aspects of predicting neurological medication efficacy using scalp electroencephalography biomarkers

Abstract

Techniques for guiding treatment of a test subject having a neurological disorder are presented. The techniques may include: producing an outlier rate data set for a test subject from a scalp electroencephalography (EEG) data set obtained from the test subject over a selected time; using the outlier rate data set to predict whether a given neurological medication (NM) will be effective in treating the test subject if the given NM is administered to the test subject to produce an NM efficacy prediction; and providing the NM efficacy prediction. Additional methods as well as related systems and computer readable media are also presented.

Description

METHODS AND RELATED ASPECTS OF PREDICTING NEUROLOGICAL MEDICATION EFFICACY USING SCALP ELECTROENCEPHALOGRAPHY BIOMARKERSCross-Reference to Related Applications

[0001] This application claims priority to, and the benefit of, U.S. Provisional Patent Application Ser. No. 63 / 666,344, filed July 1 , 2024, the disclosure of which is incorporated herein by reference.Government Funding

[0001] This invention was made with government support under grants NS132228 and NS122927 awarded by the National Institutes of Health. The government has certain rights in the invention.Field

[0002] This disclosure relates generally to therapy for neurological disorders.Background

[0003] Epilepsy is a neurological disorder characterized by disruptions in brain network organization affecting over 65 million people. Anti-seizure medications (ASMs) are the first-line treatment for epilepsy and can be highly effective: approximately two-thirds of epilepsy patients may become seizure free on ASMs. Currently, there are over 20 ASMs that are available on the market and each with its own unique mechanism of action.

[0004] Furthermore, there is no clear clinical algorithm for choosing a specific ASM over another, and an ASM is often chosen based on avoidance of potential side effects and generic epilepsy type (e.g. focal versus non-focal). ASMs are commonly adjusted throughout an individual’s treatment course due to intolerable side effects, interactions with other medications, breakthrough seizures, and / or patient preference. The clinical challenge is that there is currently no method to quantify the efficacy of ASMs and predict whether a particular ASM may or may not be effective for a specificpatient. Consequently, it can take months to years for patients to respond well to ASMs.

[0005] Accordingly, there is a need for approaches to predict and quantify the efficacy of ASMs or other neurological medications in treating subjects suffering from epilepsy or another neurological disorder.Summary

[0006] According to various embodiments, a method of guiding treatment of a test subject having a neurological disorder is presented. The method includes: producing an outlier rate data set for the test subject from a scalp electroencephalography (EEG) data set obtained from the test subject over a selected time; using the outlier rate data set to predict whether a given neurological medication (NM) will be effective in treating the test subject if the given NM is administered to the test subject to produce an NM efficacy prediction; and, providing the NM efficacy prediction, thereby guiding the treatment of the test subject having the neurological disorder. Typically, the method is computer implemented.

[0007] Various optional features of the above method embodiments include the following. The producing step may include: modeling interactions between brain regions or nodes of the test subject to generate patient-specific dynamical network models (DNMs) for the test subject using the EEG data set; computing variability of source-sink indices (VSIs), entropy of source-sink indices (ESIs), or correlates thereof, derived from the DNMs over the selected time; detecting outliers, if any, for each of the nodes from the VSIs, ESIs, or correlates thereof to produce detected outliers; and, determining one or more outlier rates from the detected outliers to produce test subject outlier rates. The using step may include comparing the test subject outlier rates with one or more control subject outlier rates. A classifier may comprise the control subject outlier rates. The EEG data set may comprise at least two EEG recordings obtained from the test subject at different time points. At least a first of the time points may comprise a time when the test subject is not being administered the given NM and at least a second of the time points may comprise a time when the test subject is being administered the given NM. The neurological disorder may comprise epilepsy. The neurological disorder may comprise schizophrenia, frontotemporal dementia (FTD), or Alzheimer’s disease. The NM may comprise an anti-seizure medication (ASM). TheNM efficacy prediction may comprise a quantitative measure related to modulation of the neurological disorder in the test subject by the given NM. The method may comprise administering the given NM to the test subject based at least in part on the NM efficacy prediction. The method may comprise discontinuing administering at least one NM to the test subject based at least in part on the NM efficacy prediction. The method may comprise providing the NM efficacy prediction to the test subject and / or to a healthcare provider. The method may comprise obtaining the EEG data set from the test subject using an electroencephalogram apparatus. A wearable device may comprise the electroencephalogram apparatus.

[0008] According to various embodiments, a system for guiding treatment of a test subject having a neurological disorder is presented. The system includes a controller, which controller comprises a processor, and a memory communicatively coupled directly or remotely to the processor, the memory storing non-transitory computer executable instructions which, when executed by the processor, perform operations comprising: producing an outlier rate data set for the test subject from a scalp electroencephalography (EEG) data set obtained from the test subject over a selected time; using the outlier rate data set to predict whether a given neurological medication (NM) will be effective in treating the test subject if the given NM is administered to the test subject to produce an NM efficacy prediction; and, providing the NM efficacy prediction.

[0009] Various optional features of the above system embodiments include the following. The system may further include an electroencephalogram apparatus configured to obtain the scalp EEG data set from the test subject, in which the controller is operably connected to the electroencephalogram apparatus and in which the non-transitory computer executable instructions which, when executed by the processor, further perform operations comprising: receiving the EEG data set obtained from the test subject. A wearable device may comprise the electroencephalogram apparatus. The non-transitory computer executable instructions which, when executed by the processor, may further perform operations comprising: modeling interactions between brain regions or nodes of the test subject to generate patient-specific dynamical network models (DNMs) for the test subject using the EEG data set; computing variability of source-sink indices (VSIs), entropy of source-sink indices (ESIs), or correlates thereof, derived from the DNMs over the selected time; detectingoutliers, if any, for each of the nodes from the VSIs, ESIs, or correlates thereof to produce detected outliers; and, determining one or more outlier rates from the detected outliers to produce test subject outlier rates. The non-transitory computer executable instructions which, when executed by the processor, may further perform operations comprising: comparing the test subject outlier rates with one or more control subject outlier rates. A classifier may comprise the control subject outlier rates. The EEG data set may comprise at least two EEG recordings obtained from the test subject at different time points. At least a first of the time points may comprise a time when the test subject is not being administered the given NM and at least a second of the time points may comprise a time when the test subject is being administered the given NM. The neurological disorder may comprise epilepsy. The neurological disorder may comprise schizophrenia, frontotemporal dementia (FTD), or Alzheimer’s disease. The NM may comprise an anti-seizure medication (ASM). The NM efficacy prediction may comprise a quantitative measure related to modulation of the neurological disorder in the test subject by the given NM.

[0010] According to various embodiments, a computer readable media is presented. The computer readable media includes non-transitory computer executable instructions which, when executed by a processor, perform operations comprising: producing an outlier rate data set for a test subject having a neurological disorder from a scalp electroencephalography (EEG) data set obtained from the test subject over a selected time; using the outlier rate data set to predict whether a given neurological medication (NM) will be effective in treating the test subject if the given NM is administered to the test subject to produce an NM efficacy prediction; and, providing the NM efficacy prediction.

[0011] Various optional features of the above system embodiments include the following. The non-transitory computer executable instructions which, when executed by the processor, may further perform operations comprising: modeling interactions between brain regions or nodes of the test subject to generate patient-specific dynamical network models (DNMs) for the test subject using the EEG data set; computing variability of source-sink indices (VSIs), entropy of source-sink indices (ESIs), or correlates thereof, derived from the DNMs over the selected time; detecting outliers, if any, for each of the nodes from the VSIs, ESIs, or correlates thereof to produce detected outliers; and, determining one or more outlier rates from thedetected outliers to produce test subject outlier rates. The non-transitory computer executable instructions which, when executed by the processor, may further perform operations comprising: comparing the test subject outlier rates with one or more control subject outlier rates. A classifier may comprise the control subject outlier rates. The neurological disorder may comprise epilepsy. The neurological disorder may comprise schizophrenia, frontotemporal dementia (FTD), or Alzheimer’s disease. The NM may comprise an anti-seizure medication (ASM). The NM efficacy prediction may comprise a quantitative measure related to modulation of the neurological disorder in the test subject by the given NM.

[0012] Combinations, (including multiple dependent combinations) of the above-described elements and those within the specification have been contemplated by the inventors and may be made, except where otherwise indicated or where contradictory.Drawings

[0013] The above and / or other aspects and advantages will become more apparent and more readily appreciated from the following detailed description of examples, taken in conjunction with the accompanying drawings, in which:

[0014] Fig. 1 is a flow chart that schematically shows exemplary method steps of guiding treatment of a test subject having a neurological disorder according to some aspects disclosed herein;

[0015] Fig. 2 is a schematic diagram of an exemplary system suitable for use with certain aspects disclosed herein;

[0016] Figs. 3A-3D are a schematic diagram of a processing pipeline from scalp EEG to classifier. (A) Network-based analysis of scalp EEG data was performed by (B) computing dynamical network models in windows of 125 milliseconds. The terms row rank (rr) and column rank (or), are sorted vectorized 1 -induced norms of the rows and columns of the state-transition matrix, A. (C) Coordinate rr and cr pairs corresponding to each channel, or node, plotted in 2-D source-sink (SS) space. Example time series of (D) SS index (SI), (E) variability of SS index (VSI), and (F) VSI outliers (VSI-O) determined via outlier detection. (G) An example table of VSI-0 rate computed per node and subsequent classification of epilepsy patients and healthy controls.

[0017] Figs. 4A-4C show aspects of the performance of an anti-seizure medication correlate. (A) Box plot of ASM ineffective, ASM effective, and Control groups. (B) Receiver operating characteristic curves for pairwise group classifications using a logistic regression classifier. (C) Topographic head maps of group-averaged z-scores of VSI outlier rates.Definitions

[0018] In order for the present disclosure to be more readily understood, certain terms are first defined below. Additional definitions for the following terms and other terms may be set forth throughout the specification. If a definition of a term set forth below is inconsistent with a definition in an application or patent that is incorporated by reference, the definition set forth in this application should be used to understand the meaning of the term.

[0019] As used in this specification and the appended claims, the singular forms“a,” “an,” and “the” include plural references unless the context clearly dictates otherwise. Thus, for example, a reference to “a method” includes one or more methods, and / or steps of the type described herein and / or which will become apparent to those persons skilled in the art upon reading this disclosure and so forth.

[0020] It is also to be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting. Further, unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure pertains. In describing and claiming the methods, systems, and computer readable media, the following terminology, and grammatical variants thereof, will be used in accordance with the definitions set forth below.

[0021] Classifier. As used herein, “classifier” generally refers to algorithm computer code that receives, as input, test data and produces, as output, a classification of the input data as belonging to one or another class.

[0022] Data set: As used herein, “data set” refers to a group or collection of information, values, or data points related to or associated with one or more objects, records, and / or variables. In some embodiments, a given data set is organized as, or included as part of, a matrix or tabular data structure. In some embodiments, a data set is encoded as a feature vector corresponding to a given object, record, and / orvariable, such as a given test or reference subject. For example, a medical data set for a given subject can include one or more observed values of one or more variables associated with that subject.

[0023] Electronic neural network. As used herein, “electronic neural network” refers to a machine learning algorithm or model that includes layers of at least partially interconnected artificial neurons (e.g., perceptrons or nodes) organized as input and output layers with one or more intervening hidden layers that together form a network that is or can be trained to classify data, such as test subject medical data sets (e.g., medical images or the like).

[0024] Machine Learning Algorithm: As used herein, "machine learning algorithm" generally refers to an algorithm, executed by computer, that automates analytical model building, e.g., for clustering, classification or pattern recognition. Machine learning algorithms may be supervised or unsupervised. Learning algorithms include, for example, artificial neural networks (e.g., back propagation networks, transformer networks, etc.), discriminant analyses (e.g., Bayesian classifier or Fisher’s analysis), multiple-instance learning (MIL), support vector machines, decision trees (e.g., recursive partitioning processes such as CART -classification and regression trees, or random forests), linear classifiers (e.g., multiple linear regression (MLR), partial least squares (PLS) regression, and principal components regression), hierarchical clustering, and cluster analysis. A dataset on which a machine learning algorithm learns can be referred to as "training data." A model produced using a machine learning algorithm is generally referred to herein as a “machine learning model.”

[0025] Subject: As used herein, “subject” or “test subject” refers to an animal, such as a mammalian species (e.g., human) or avian (e.g., bird) species. More specifically, a subject can be a vertebrate, e.g., a mammal such as a mouse, a primate, a simian or a human. Animals include farm animals (e.g., production cattle, dairy cattle, poultry, horses, pigs, and the like), sport animals, and companion animals (e.g., pets or support animals). A subject can be a healthy individual, an individual that has or is suspected of having a disease or pathology or a predisposition to the disease or pathology, or an individual that is in need of therapy or suspected of needing therapy. The terms “individual” or “patient” are intended to be interchangeable with “subject.” A“reference subject” refers to a subject known to have or lack specific properties (e.g., a known pathology, such as melanoma and / or the like).

[0026] Value: As used herein, “value” generally refers to an entry in a dataset that can be anything that characterizes the feature to which the value refers. This includes, without limitation, numbers, words or phrases, symbols (e.g., + or -) or degrees.Description of the Embodiments

[0027] Reference will now be made in detail to example implementations. These embodiments are described in sufficient detail to enable those skilled in the art to practice the invention and it is to be understood that other embodiments may be utilized and that changes may be made without departing from the scope of the invention. The following description is, therefore, merely exemplary.

[0028] I. Introduction

[0029] Neurological medications, such as an anti-seizure medication (ASM) regimen that relieves seizure burden for patients with epilepsy can take weeks to months to establish or become ineffective over time and require revision. There is no clear clinical algorithm for choosing one specific ASM regimen over another, and an ASM is often chosen based on a patient’s individual characteristics and electroencephalogram (EEG). These approaches can vary from patient to patient and may be prohibitive in low-resource settings. In the present disclosure, we present, for example, an objective, quantitative, and automated approach to measure ASM efficacy utilizing scalp EEG from three groups of adults: epilepsy patients on ineffective ASMs, epilepsy patients on effective ASMs, and healthy controls. For each patient, we build a dynamical network model (DNM) from resting state EEG that estimates regional interactions over time using a novel source-sink (SS) index. We demonstrate that patients with ineffective ASMs exhibit more abrupt changes in the SS index throughout a scalp EEG than patients with effective ASMs or healthy controls. Accordingly, in some embodiments, the process disclosed herein may provide a quantitative correlate for ASMs that can be used to compare regimens for individual patients.

[0030] To illustrate, Fig. 1 is a flow chart that schematically shows exemplary method steps of guiding treatment of a test subject having a neurological disorder according to some aspects disclosed herein. As shown, method 100 includes producing an outlier rate data set for the test subject from a scalp electroencephalography (EEG) data set obtained from the test subject over a selected time (step 102). Method 100 also includes using the outlier rate data set to predict whether a given neurological medication (NM) will be effective in treating the test subject if the given NM is administered to the test subject to produce an NM efficacy prediction (step 104). In addition, method 100 also includes providing the NM efficacy prediction (step 106).

[0031] In some embodiments, the producing step of method 100 includes modeling interactions between brain regions or nodes of the test subject to generate patient-specific dynamical network models (DNMs) for the test subject using the EEG data set; computing variability of source-sink indices (VSIs), entropy of source-sink indices (ESIs), or correlates thereof, derived from the DNMs over the selected time; detecting outliers, if any, for each of the nodes from the VSIs, ESIs, or correlates thereof to produce detected outliers; and, determining one or more outlier rates from the detected outliers to produce test subject outlier rates. In some embodiments, the using step of method 100 comparing the test subject outlier rates with one or more control subject outlier rates. In some embodiments, a classifier comprises the control subject outlier rates.

[0032] In some embodiments, the EEG data set comprises at least two EEG recordings obtained from the test subject at different time points. In some embodiments, at least a first of the time points comprises a time when the test subject is not being administered the given NM and at least a second of the time points comprises a time when the test subject is being administered the given NM.

[0033] The methods and related aspects of the present disclosure can be used to guide treatment of various neurological disorders. In some embodiments, for example, the neurological disorder comprises epilepsy. In other exemplary embodiments, the neurological disorder comprises schizophrenia, frontotemporal dementia (FTD), or Alzheimer’s disease.

[0034] In some embodiments, the NM comprises an anti-seizure medication (ASM). In some embodiments, the NM efficacy prediction comprises a quantitativemeasure related to modulation of the neurological disorder in the test subject by the given NM. In some embodiments, method 100 includes administering the given NM to the test subject based at least in part on the NM efficacy prediction. In some embodiments, method 100 includes discontinuing administering at least one NM to the test subject based at least in part on the NM efficacy prediction. In some embodiments, method 100 includes providing the NM efficacy prediction to the test subject and / or to a healthcare provider. In some embodiments, method 100 includes obtaining the EEG data set from the test subject using an electroencephalogram apparatus. In some embodiments, a wearable device comprises the electroencephalogram apparatus.

[0035] Fig. 2 is a schematic diagram of a hardware computer system 200 suitable for implementing various embodiments. For example, Fig. 2 illustrates various hardware, software, and other resources that can be used in implementations of any of methods disclosed herein, including, e.g., method 100 and / or one or more instances of an electronic neural network. System 200 includes training corpus source 202 and computer 201 . Training corpus source 202 and computer 201 may be communicatively coupled by way of one or more networks 204, e.g., the internet.

[0036] Training corpus source 202 may include an electronic clinical records system, such as an LIS, a database, a compendium of clinical data, or any other source of images suitable for use as a training corpus as disclosed herein. Due to hardware volatile memory storage limitations, each constituent image of an image may be broken down into a number of tiles, which may be, e.g., 128 pixels by 128 pixels. Such tiles are examples of “components” as that term is used herein. According to some embodiments, each component is implemented as a vector, such as a feature vector, that represents a respective tile. Thus, the term “component” refers to both a tile and a feature vector representing a tile.

[0037] Computer 201 may be implemented as any of a desktop computer, a laptop computer, can be incorporated in one or more servers, clusters, or other computers or hardware resources, or can be implemented using cloud-based resources. Computer 201 includes volatile memory 214 and persistent memory 212, the latter of which can store computer-readable instructions, that, when executed by electronic processor 210, configure computer 201 to perform methods disclosed herein, including method 100, and / or form or store any electronic neural network, and / or perform any classification technique as described herein. Computer 201 furtherincludes network interface 208, which communicatively couples computer 201 to training corpus source 202 via network 204. Other configurations of system 200, associated network connections, and other hardware, software, and service resources are possible. As shown, system 200 also includes electroencephalogram apparatus 218 disposed on test subject 216 and operably connected to system 200 via network 204. Electroencephalogram apparatus 218 is typically used to obtain the EEG data set from test subject 216.

[0038] Certain embodiments can be performed using a computer program or set of programs. The computer programs can exist in a variety of forms both active and inactive. For example, the computer programs can exist as software program(s) comprised of program instructions in source code, object code, executable code or other formats; firmware program(s), or hardware description language (HDL) files. Any of the above can be embodied on a transitory or non-transitory computer readable medium, which include storage devices and signals, in compressed or uncompressed form. Exemplary computer readable storage devices include conventional computer system RAM (random access memory), ROM (read-only memory), EPROM (erasable, programmable ROM), EEPROM (electrically erasable, programmable ROM), and magnetic or optical disks or tapes.

[0039] II. Description of Example Embodiments

[0040] EXAMPLE: Scalp EEG Correlates of Anti-Seizure Medications inAdult Epilepsy

[0041] 1. Introduction

[0042] In this example, we introduce an objective, quantitative, and automated approach to measure ASM efficacy by utilizing scalp EEG from adult non-intractable (ASM-responding) epilepsy patients who underwent routine scalp EEGs for ASM adjustments. Using each EEG, we model interactions between brain regions, or nodes, by constructing patient-specific dynamical network models (DNMs). DNMs are generative, linear time-varying (LTV) models that capture the dynamics of nodal interactions in the EEG network. We hypothesize that an epileptic cortical network fluctuates more with ineffective ASMs than with effective ASMs, suggesting shifts of nodal balance. We further hypothesize that nodal interactions remain relativelystationary in healthy controls, suggesting that effective ASMs restore nodal balance in the EEG network.

[0043] To test our hypotheses, we compute a metric that captures the variability of the source-sink (SS) index derived from each DNM. We demonstrate that our process provides a quantitative correlate for ASMs that can be used to compare regimens for individual patients.

[0044] 2. Methods

[0045] A. Dataset of Epileptic Patients and Healthy Controls

[0046] The scalp EEG recordings in this work originate from 7 non-intractable epilepsy patients of the Johns Hopkins Outpatient Center (JHOC) and 7 healthy control participants (Control group) of the open access Nencki-Symfonia database. For the epilepsy patients, an EEG recording was collected while the patient was ASM naive or on ineffective ASMs (ASM ineff. group) and another recording was collected after at least 6 months of seizure freedom on effective ASMs (ASM eff. group). Raw scalp EEGs from JHOC were recorded with a 40-channel EEG-1100 system (Nihon Kohden America, Irvine, CA, USA), which amplified and sampled the data at 200 Hz. The use of de-identified scalp EEGs and pertinent medical record information from JHOC in this study was approved by the Johns Hopkins Medicine Institutional Review Board. EEG recordings from the Nencki-Symfonia database were collected at 1000 Hertz in a 128-channel arrangement. Table I outlines the demographic characteristics of the populations studied.TABLE IDEMOGRAPHIC CHARACTERISTICS OF EPILEPSY PATIENTS AND HEALTHY CONTROLSMean age, years (± D) 17,80 (6.21) 20.21 (5.31) 22.26 (4.30) Female sex, n (%) 3 (43) - 3 (43) Mean recording length, minutes (±SD) 25.7 (3.22) 26.03 (2.33) 12.01 (0.22)Mean age at epilepsy onset, years (±£>D) 14.23 (10.01) -Mean duration of seizure freedom, years (±£>D) 3.51 (2.31)

[0047] B. Pre-processing of EEG Recordings

[0048] Initially, each recording was standardized to the 10-20 montage of 19EEG channels. Then, power line rejection at 50 or 60 Hertz was performed via asecond order Butterworth band pass filter. In addition, a fourth order Butterworth bandpass filter with upper and lower cutoff frequencies of 1 and 90 Hertz was applied. Next, independent component analysis (ICA) for artifact removal (muscle, eye movements) in windows of 90 seconds was performed via the EEGLab ICA toolbox in MATLAB. Following this, each EEG recording was down sampled to 200 Hertz before further computations.

[0049] C. Estimating Dynamical Network Models from EEG

[0050] An EEG recording is a time series that is denoted, whereN is the number of EEG channels (nodes in the network). For each 125 milliseconds of EEG, we then construct the following linear time-invariant model describing how x(t + 1) = cc(t) (1) wis the state-transition matrix used to determine the next time point (millisecond) of(Fig. 3A). We form patient-specific DNMs by concatenating the LTI systems of (1 ) over time to form an LTV. These generative LTV systems capture the dynamic influence between brain regions from the original EEG recording.

[0051] The A matrices in (1 ) for the DNM at each time point of the EEG recording were estimated via the least squares solution to the over-determined system of linear equations in Fig. 3B. An entryof the matrix corresponds to the effect of node on node Similarly, therow corresponds to the incoming influence of the, 'th network on node and the 3 column corresponds to the outgoing influence of node 3 on the network.

[0052] D. Computing the Source-Sink Index

[0053] The SS index is a network-based biomarker of epileptogenicity validated in intracranial EEG which can identify nodes as sinks or sources in a cortical network: a sink is highly influenced by the network while a source is highly influential on the network.

[0054] For an epileptic cortical network, sinks correspond to volatile, epileptogenic regions that are inhibited by the network to suppress seizures when at rest. The degree to which each node is a source or sink can be derived from the statetransition matrices A. Determination of sources and sinks as depicted in Fig. 3A-B is via ranking of the sorted, vectorized 1 -norms of the rows and columns in A to generatea row rank and column rank corresponding to each node. A node’s rr and form a pair in the coordinate system of SS space. Fig. 3C is a schematic of SS space with example plottedrT>cr^ coordinates. The SS space facilitates the derivation of the following relationships:whereJ correspond to the ideal sink and the ideal source respectively in SS space, i corresponds to the node,measures the influence by sources on node and conm measures the connectivity of sinks to node i. The SS index is a measure of epileptogenicity for each node over each A matrix of the DNM and is returned by combining (2), (4), and (5):SSi = sinki * infli * eonni (6)

[0055] E. Computing Variability of Source-Sink Index

[0056] We analyzed the SS index fluctuations over time by computing its variance, or variability. Previous studies have highlighted variance as a notable feature of EEG signals for epileptiform channel selection or for classification tasks.

[0057] We computed the variability of SS indices (VSI) at each node over time as:where w corresponds to a windowed time segment of SS indices. Lower variability within a window relates to negligible changes in network dynamics and a more even distribution of SS indices. In contrast, higher variability relates to more extreme changes in network dynamics and a highly dispersed distribution of SS indices.

[0058] For an epileptic cortical network, the SS index over time is characterized by a baseline of low-amplitude activity punctuated by transient high-amplitude activity.Similarly, the variability of the SS index (VSI) over time exhibits outliers or transient extreme increases in amplitude.

[0059] F. Detecting Outliers of Source-Sink Index Variability

[0060] We performed outlier detection on each node of the VSI time series using median absolute deviation (MAD):is the median of the time series at node7j>iis the modified z-score at each time point of nodeand ® is the normal cumulative distribution function. The 4 " as in (9) is the quantile function for a normal distribution. Outliers were determined from the modified z-scores via the application of a threshold set to 3.5 standard deviations from the median.

[0061] G. Computing Outlier Rate as Anti-Seizure Medication Correlate

[0062] We standardized the number of outliers of each node and the time length of each recording by generating a VS I outlier rate in units of outliers per second. Thus, the VS I outlier rate is produced as our correlate for ASM efficacy. We hypothesized that the VSI outlier rates of patients with epilepsy will decrease when on effective ASMs and will resemble the VSI outlier rates of healthy controls.

[0063] H. Classifying Patients Based on Outlier Rate

[0064] The three groups (ASM ineff. , ASM eff. , and Control) were classified using the outlier rate as a single independent variable via logistic regression using a leave-one-out cross-validation technique with 10 repetitions in Python. The receiveroperator characteristic curves were generated from aggregated true and predicted label probabilities for each test case.

[0065] 3. Results

[0066] To test our hypothesis, we determined the rates of VSI outliers for each individual. We present the results of our ASM regimen correlate for (a) distinguishing among EEGs corresponding to ASM ineff., ASM eff., and Control groups; (b) sensitivity and specificity for each pairwise group comparison; and (c) visualization of VSI outlier rates via topographic head maps.

[0067] We observed means of 4.65 VSI outliers per second (SD = 3.79), 0.64 VSI outliers per second (SD = 0.74), and 0.73 VSI outliers per second (SD = 0.40) for the ASM ineff. , ASM eff. , and Control groups, respectively. The three tests of the a priori hypothesis were conducted at a Bonferroni-adjusted alpha level of 0.0166 (0.05 / 3). A Mann-Whitney U test showed that the VSI outlier rates of the ASM ineff. Group were significantly higher (U = 49.0, P < .001 ) than those of the Control group (Fig. 4A). The pairwise comparisons of VSI outlier rates for the ASM ineff. and ASM eff. groups as well as the ASM eff. and the Control groups were not statistically significant.

[0068] The receiver operating curves for the pairwise group classifications in Fig. 4B showed areas under the curve (AUC) of 0.70, 0.68, and 0.41 for ASM ineff. vs. Control, ASM ineff. vs. ASM eff., and ASM eff. vs Control, respectively. The sensitivity and specificity associated with each pairwise group classification were 0.96 and 0.71 , 0.75 and 0.97, and 0.08 and 0.0, respectively. A visualization of the VSI outlier rate across groups in the form of topographic head maps as in Fig. 4C depicts regional and global differences. Visually, distinct regional differences in VSI outlier rate were greatest for the ASM ineff. group and the least for the Control group.

[0069] 4. Discussion

[0070] In this example, we hypothesized that the VSI outlier rates of patients with epilepsy will decrease when on effective ASMs and will resemble those of healthy controls. To test this hypothesis, we computed the VSI outlier rates of epilepsy patients and healthy controls and applied this metric to group comparisons. One of three pairwise group comparisons was statistically significant (ASM ineff. vs Control). The non-significant pairwise group comparison ASM eff. vs. Control is of interest and suggests epileptogenicity reduction mediated by effective ASMs can be computed by network-based approaches such as our VSI outlier rate. The remaining nonsignificant pairwise comparison between ASM ineff. and ASM eff. poses as an area of further exploration of features, such as epilepsy type. However, the trend shows that the VSI outlier rate is lower when patients are on effective ASMs.

[0071] AUCs of the pairwise group classifications were below 0.8 with one notable AUC for ASM eff. vs Control that was less than chance. In accordance with our hypothesis, difficulties of classification for these two groups highlights a resemblance of their VSI outlier rates. The greatest sensitivity was for the ASM ineff.vs Control groups classification and the greatest specificity was for the ASM ineff. vs ASM eff. groups classification. The latter classifier may help identify usage of ineffective ASMs at a low false positive rate.

[0072] 5. Conclusion

[0073] In this example, we demonstrate for the first time the utility of computing VSI outlier rate on scalp EEG as a correlate of ASM efficacy. Patients with epilepsy who remained seizure free on ASMs had VSI outlier rates that were closer to those of healthy controls, therefore indicating a shift to more balanced network dynamics. Our VSI outlier rate approach provides a quantitative and objective ASM correlate and enables comparison of ASMs. Its role as an automated method is to support the personalized treatment of epilepsy. In the future, we would like to proactively select effective ASMs, expand our dataset, pursue the discovery of additional features, and develop models of ASM-related network interactions.

[0074] Some further aspects are defined in the following clauses:

[0075] Clause 1 : A method of guiding treatment of a test subject having a neurological disorder, the method comprising: producing an outlier rate data set for the test subject from a scalp electroencephalography (EEG) data set obtained from the test subject over a selected time; using the outlier rate data set to predict whether a given neurological medication (NM) will be effective in treating the test subject if the given NM is administered to the test subject to produce an NM efficacy prediction; and, providing the NM efficacy prediction, thereby guiding the treatment of the test subject having the neurological disorder.

[0076] Clause 2: The method of Clause 1 , wherein the producing step comprises: modeling interactions between brain regions or nodes of the test subject to generate patient-specific dynamical network models (DNMs) for the test subject using the EEG data set; computing variability of source-sink indices (VSIs), entropy of source-sink indices (ESIs), or correlates thereof, derived from the DNMs over the selected time; detecting outliers, if any, for each of the nodes from the VSIs, ESIs, or correlates thereof to produce detected outliers; and, determining one or more outlier rates from the detected outliers to produce test subject outlier rates.

[0077] Clause 3: The method of Clause 1 or Clause 2, wherein the using step comprises comparing the test subject outlier rates with one or more control subject outlier rates.

[0078] Clause 4: The method of any one of the preceding Clauses 1 -3, wherein a classifier comprises the control subject outlier rates.

[0079] Clause 5: The method of any one of the preceding Clauses 1 -4, wherein the EEG data set comprises at least two EEG recordings obtained from the test subject at different time points.

[0080] Clause 6: The method of any one of the preceding Clauses 1 -5, wherein at least a first of the time points comprises a time when the test subject is not being administered the given NM and at least a second of the time points comprises a time when the test subject is being administered the given NM.

[0081] Clause 7: The method of any one of the preceding Clauses 1 -6, wherein the neurological disorder comprises epilepsy.

[0082] Clause 8: The method of any one of the preceding Clauses 1 -7, wherein the neurological disorder comprises schizophrenia, frontotemporal dementia (FTD), or Alzheimer’s disease.

[0083] Clause 9: The method of any one of the preceding Clauses 1 -8, wherein the NM comprises an anti-seizure medication (ASM).

[0084] Clause 10: The method of any one of the preceding Clauses 1 -9, wherein the NM efficacy prediction comprises a quantitative measure related to modulation of the neurological disorder in the test subject by the given NM.

[0085] Clause 11 : The method of any one of the preceding Clauses 1 -10, comprising administering the given NM to the test subject based at least in part on the NM efficacy prediction.

[0086] Clause 12: The method of any one of the preceding Clauses 1 -11 , comprising discontinuing administering at least one NM to the test subject based at least in part on the NM efficacy prediction.

[0087] Clause 13: The method of any one of the preceding Clauses 1 -12, comprising providing the NM efficacy prediction to the test subject and / or to a healthcare provider.

[0088] Clause 14: The method of any one of the preceding Clauses 1 -13, comprising obtaining the EEG data set from the test subject using an electroencephalogram apparatus.

[0089] Clause 15: The method of any one of the preceding Clauses 1 -14, wherein a wearable device comprises the electroencephalogram apparatus.

[0090] Clause 16: A system for guiding treatment of a test subject having a neurological disorder, comprising: a controller, which controller comprises a processor, and a memory communicatively coupled directly or remotely to the processor, the memory storing non-transitory computer executable instructions which, when executed by the processor, perform operations comprising: producing an outlier rate data set for the test subject from a scalp electroencephalography (EEG) data set obtained from the test subject over a selected time; using the outlier rate data set to predict whether a given neurological medication (NM) will be effective in treating the test subject if the given NM is administered to the test subject to produce an NM efficacy prediction; and, providing the NM efficacy prediction.

[0091] Clause 17: The system of Clause 16, further comprising an electroencephalogram apparatus configured to obtain the scalp EEG data set from the test subject, wherein the controller is operably connected to the electroencephalogram apparatus and wherein the non-transitory computer executable instructions which, when executed by the processor, further perform operations comprising: receiving the EEG data set obtained from the test subject.

[0092] Clause 18: The system of Clause 16 or Clause 17, wherein a wearable device comprises the electroencephalogram apparatus.

[0093] Clause 19: The system of any one of the preceding Clauses 16-18, wherein the non-transitory computer executable instructions which, when executed by the processor, further perform operations comprising: modeling interactions between brain regions or nodes of the test subject to generate patient-specific dynamical network models (DNMs) for the test subject using the EEG data set; computing variability of source-sink indices (VSIs), entropy of source-sink indices (ESIs), or correlates thereof, derived from the DNMs over the selected time; detecting outliers, if any, for each of the nodes from the VSIs, ESIs, or correlates thereof to produce detected outliers; and, determining one or more outlier rates from the detected outliers to produce test subject outlier rates.

[0094] Clause 20: The system of any one of the preceding Clauses 16-19, wherein the non-transitory computer executable instructions which, when executed by the processor, further perform operations comprising: comparing the test subject outlier rates with one or more control subject outlier rates.

[0095] Clause 21 : The system of any one of the preceding Clauses 16-20, wherein a classifier comprises the control subject outlier rates.

[0096] Clause 22: The system of any one of the preceding Clauses 16-21 , wherein the EEG data set comprises at least two EEG recordings obtained from the test subject at different time points.

[0097] Clause 23: The system of any one of the preceding Clauses 16-22, wherein at least a first of the time points comprises a time when the test subject is not being administered the given NM and at least a second of the time points comprises a time when the test subject is being administered the given NM.

[0098] Clause 24: The system of any one of the preceding Clauses 16-23, wherein the neurological disorder comprises epilepsy.

[0099] Clause 25: The system of any one of the preceding Clauses 16-24, wherein the neurological disorder comprises schizophrenia, frontotemporal dementia (FTD), or Alzheimer’s disease.

[0100] Clause 26: The system of any one of the preceding Clauses 16-25, wherein the NM comprises an anti-seizure medication (ASM).

[0101] Clause 27: The system of any one of the preceding Clauses 16-26, wherein the NM efficacy prediction comprises a quantitative measure related to modulation of the neurological disorder in the test subject by the given NM.

[0102] Clause 28: A computer readable media comprising non-transitory computer executable instructions which, when executed by at least electronic processor, perform at least: producing an outlier rate data set for a test subject having a neurological disorder from a scalp electroencephalography (EEG) data set obtained from the test subject over a selected time; using the outlier rate data set to predict whether a given neurological medication (NM) will be effective in treating the test subject if the given NM is administered to the test subject to produce an NM efficacy prediction; and, providing the NM efficacy prediction.

[0103] Clause 29: The computer readable media of Clause 28, wherein the non- transitory computer executable instructions which, when executed by the processor, further perform operations comprising: modeling interactions between brain regions or nodes of the test subject to generate patient-specific dynamical network models (DNMs) for the test subject using the EEG data set; com puting variability of sourcesink indices (VSIs), entropy of source-sink indices (ESIs), or correlates thereof,derived from the DNMs over the selected time; detecting outliers, if any, for each of the nodes from the VSIs, ESIs, or correlates thereof to produce detected outliers; and, determining one or more outlier rates from the detected outliers to produce test subject outlier rates.

[0104] Clause 30: The computer readable media of Clause 28 or Clause 29, wherein the non-transitory computer executable instructions which, when executed by the processor, further perform operations comprising: comparing the test subject outlier rates with one or more control subject outlier rates.

[0105] Clause 31 : The computer readable media of any one of the preceding Clauses 28-30, wherein a classifier comprises the control subject outlier rates.

[0106] Clause 32: The computer readable media of any one of the preceding Clauses 28-31 , wherein the neurological disorder comprises epilepsy.

[0107] Clause 33: The computer readable media of any one of the preceding Clauses 28-32, wherein the neurological disorder comprises schizophrenia, frontotemporal dementia (FTD), or Alzheimer’s disease.

[0108] Clause 34: The computer readable media of any one of the preceding Clauses 28-33, wherein the NM comprises an anti-seizure medication (ASM).

[0109] Clause 35: The computer readable media of any one of the preceding Clauses 28-34, wherein the NM efficacy prediction comprises a quantitative measure related to modulation of the neurological disorder in the test subject by the given NM.

[0110] While the invention has been described with reference to the exemplary embodiments thereof, those skilled in the art will be able to make various modifications to the described embodiments without departing from the true spirit and scope. The terms and descriptions used herein are set forth by way of illustration only and are not meant as limitations. In particular, although the method has been described by examples, the steps of the method can be performed in a different order than illustrated or simultaneously. Those skilled in the art will recognize that these and other variations are possible within the spirit and scope as defined in the following claims and their equivalents. All patents, patent applications, other publications or documents, and the like cited herein are incorporated by reference in their entirety for all purposes to the same extent as if each individual item were specifically and individually indicated to be so incorporated by reference.

Claims

What is claimed is:1 . A method of guiding treatment of a test subject having a neurological disorder, the method comprising: producing an outlier rate data set for the test subject from a scalp electroencephalography (EEG) data set obtained from the test subject over a selected time; using the outlier rate data set to predict whether a given neurological medication (NM) will be effective in treating the test subject if the given NM is administered to the test subject to produce an NM efficacy prediction; and, providing the NM efficacy prediction, thereby guiding the treatment of the test subject having the neurological disorder.

2. The method of claim 1 , wherein the producing step comprises: modeling interactions between brain regions or nodes of the test subject to generate patient-specific dynamical network models (DNMs) for the test subject using the EEG data set; computing variability of source-sink indices (VSIs), entropy of source-sink indices (ESIs), or correlates thereof, derived from the DNMs over the selected time; detecting outliers, if any, for each of the nodes from the VSIs, ESIs, or correlates thereof to produce detected outliers; and, determining one or more outlier rates from the detected outliers to produce test subject outlier rates.

3. The method of claim 2, wherein the using step comprises comparing the test subject outlier rates with one or more control subject outlier rates.

4. The method of claim 3, wherein a classifier comprises the control subject outlier rates.

5. The method of claim 1 , wherein the EEG data set comprises at least two EEG recordings obtained from the test subject at different time points.

6. The method of claim 5, wherein at least a first of the time points comprises a time when the test subject is not being administered the given NM and at least a second of the time points comprises a time when the test subject is being administered the given NM.

7. The method of claim 1 , wherein the neurological disorder comprises epilepsy.

8. The method of claim 1 , wherein the neurological disorder comprises schizophrenia, frontotemporal dementia (FTD), or Alzheimer’s disease.

9. The method of claim 1 , wherein the NM comprises an anti-seizure medication (ASM).

10. The method of claim 1 , wherein the NM efficacy prediction comprises a quantitative measure related to modulation of the neurological disorder in the test subject by the given NM.11 . The method of claim 1 , comprising administering the given NM to the test subject based at least in part on the NM efficacy prediction.

12. The method of claim 1 , comprising discontinuing administering at least one NM to the test subject based at least in part on the NM efficacy prediction.

13. The method of claim 1 , comprising providing the NM efficacy prediction to the test subject and / or to a healthcare provider.

14. The method of claim 1 , comprising obtaining the EEG data set from the test subject using an electroencephalogram apparatus.

15. The method of claim 14, wherein a wearable device comprises the electroencephalogram apparatus.

16. A system for guiding treatment of a test subject having a neurological disorder, the system comprising: a controller, which controller comprises a processor, and a memory communicatively coupled directly or remotely to the processor, the memory storing non-transitory computer executable instructions which, when executed by the processor, perform operations comprising: producing an outlier rate data set for the test subject from a scalp electroencephalography (EEG) data set obtained from the test subject over a selected time; using the outlier rate data set to predict whether a given neurological medication (NM) will be effective in treating the test subject if the given NM is administered to the test subject to produce an NM efficacy prediction; and, providing the NM efficacy prediction.

17. The system of claim 16, further comprising an electroencephalogram apparatus configured to obtain the scalp EEG data set from the test subject, wherein the controller is operably connected to the electroencephalogram apparatus and wherein the non-transitory computer executable instructions which, when executed by the processor, further perform operations comprising: receiving the EEG data set obtained from the test subject.

18. The system of claim 17, wherein a wearable device comprises the electroencephalogram apparatus.

19. The system of claim 16, wherein the non-transitory computer executable instructions which, when executed by the processor, further perform operations comprising: modeling interactions between brain regions or nodes of the test subject to generate patient-specific dynamical network models (DNMs) for the test subject using the EEG data set; computing variability of source-sink indices (VSIs), entropy of source-sink indices (ESIs), or correlates thereof, derived from the DNMs over the selected time; detecting outliers, if any, for each of the nodes from the VSIs, ESIs, orcorrelates thereof to produce detected outliers; and, determining one or more outlier rates from the detected outliers to produce test subject outlier rates.

20. The system of claim 19, wherein the non-transitory computer executable instructions which, when executed by the processor, further perform operations comprising: comparing the test subject outlier rates with one or more control subject outlier rates.21 . The system of claim 20, wherein a classifier comprises the control subject outlier rates.

22. The system of claim 16, wherein the EEG data set comprises at least two EEG recordings obtained from the test subject at different time points.

23. The system of claim 22, wherein at least a first of the time points comprises a time when the test subject is not being administered the given NM and at least a second of the time points comprises a time when the test subject is being administered the given NM.

24. The system of claim 16, wherein the neurological disorder comprises epilepsy.

25. The system of claim 16, wherein the neurological disorder comprises schizophrenia, frontotemporal dementia (FTD), or Alzheimer’s disease.

26. The system of claim 16, wherein the NM comprises an anti-seizure medication (ASM).

27. The system of claim 16, wherein the NM efficacy prediction comprises a quantitative measure related to modulation of the neurological disorder in the test subject by the given NM.

28. A computer readable media, comprising non-transitory computer executable instructions which, when executed by a processor, perform operations comprising: producing an outlier rate data set for a test subject having a neurological disorder from a scalp electroencephalography (EEG) data set obtained from the test subject over a selected time; using the outlier rate data set to predict whether a given neurological medication (NM) will be effective in treating the test subject if the given NM is administered to the test subject to produce an NM efficacy prediction; and, providing the NM efficacy prediction.

29. The computer readable media of claim 28, wherein the non-transitory computer executable instructions which, when executed by the processor, further perform operations comprising: modeling interactions between brain regions or nodes of the test subject to generate patient-specific dynamical network models (DNMs) for the test subject using the EEG data set; computing variability of source-sink indices (VSIs), entropy of source-sink indices (ESIs), or correlates thereof, derived from the DNMs over the selected time; detecting outliers, if any, for each of the nodes from the VSIs, ESIs, or correlates thereof to produce detected outliers; and, determining one or more outlier rates from the detected outliers to produce test subject outlier rates.

30. The computer readable media of claim 29, wherein the non-transitory computer executable instructions which, when executed by the processor, further perform operations comprising: comparing the test subject outlier rates with one or more control subject outlier rates.31 . The computer readable media of claim 30, wherein a classifier comprises the control subject outlier rates.

32. The computer readable media of claim 28, wherein the neurological disorder comprises epilepsy.

33. The computer readable media of claim 28, wherein the neurological disorder comprises schizophrenia, frontotemporal dementia (FTD), or Alzheimer’s disease.

34. The computer readable media of claim 28, wherein the NM comprises an anti-seizure medication (ASM).

35. The computer readable media of claim 28, wherein the NM efficacy prediction comprises a quantitative measure related to modulation of the neurological disorder in the test subject by the given NM.

Images