Methods for identifying subjects for inclusion and / or exclusion in clinical trials

Abstract

Methods for selecting candidate subjects based on predicted responses to experimental treatments or therapeutic agents, and related methods for identifying subjects eligible to participate in clinical trials who may be likely to respond to placebo, are disclosed herein. The methods disclosed herein are useful for identifying and / or selecting subjects for inclusion in or exclusion from clinical trials.

Description

Technical Field

[0001] (Cross-Reference to Related Applications) This application claims the benefit of U.S. Provisional Patent Application No. 63 / 533,888, filed on August 21, 2023. The entire disclosure of the above application is incorporated herein by reference.

[0002] (Field of the Invention) The present invention relates to methods for identifying subjects for inclusion and / or exclusion in clinical trials based on the determined likelihood of a subject to respond to a drug and / or placebo. Specifically, the present invention relates to the development, training, and application of a set of algorithms that enable quantitative selection of individual subjects for inclusion in central nervous system (CNS) clinical trials targeting specific neural targets. Background Art

[0003] (Background) In general, subjects participating in clinical trials for experimental therapies are selected based on inclusion criteria that help define subjects matching the patient population that the trial is intended to study, and exclusion criteria that exclude subjects for whom the experimental therapy may be unsafe or ineffective. These inclusion and exclusion criteria are established based on several factors including, for example, signs or symptoms that the experimental therapy (e.g., a drug) is intended to treat or address, the outcome of interest, regulatory guidelines, and prior evidence including available safety and efficacy data obtained from preclinical studies and previous clinical trials.

[0004] While such conventional information sources are useful for defining target populations for experimental therapies, they experience and are susceptible to several shortcomings. For example, there is a lack of available techniques to identify individual subjects likely to respond to experimental therapies in clinical trials based on the predicted likelihood of a response to a specific neural target. Currently, there are no quantitative methods that use functional neuronal information to select or exclude subjects from participation in clinical trials based on the predicted probability of responding to experimental therapy.

[0005] (Description of related technologies) CNS drug development fails at a high rate, partly due to the significant heterogeneity within CNS diagnostic categories. For example, patients with the same diagnosis do not always share the same underlying pathology, resulting in a high rate of non-responders or partial responders to administered treatments and medications. These challenges are exacerbated because many clinical trials for novel experimental treatments for CNS conditions are designed to enroll subjects who meet specific behavioral diagnostic criteria without any method for distinguishing between potential subgroups or subpopulations between those likely to respond to the experimental treatment and those unlikely to respond. Currently, there is no method to optimize subject enrollment in clinical trials based on the predicted likelihood of such subjects responding to experimental treatments (e.g., drugs).

[0006] There is a growing effort to incorporate neurobiological markers and / or neuroimaging markers into the development and final formulation of novel CNS medicines. For example, recanemab, prescribed to slow the progression of Alzheimer's disease, requires confirmation of the presence of amyloid-beta plaques in the patient's brain via cerebrospinal fluid samples or positron emission tomography (PET) scans before prescription, and regular magnetic resonance imaging (MRI) scans of the patient for safety monitoring during treatment. However, an approach that requires neuroscans or lumbar punctures in all potential patients as a prerequisite for treatment is costly, invasive, and impractical.

[0007] Understanding how the manifestation of various symptoms relates to the biological mechanisms and circuits within the human brain is crucial. This is greatly supported by non-invasive imaging techniques such as MRI, which have mapped neural functions related to a wide range of behaviors, including psychiatric symptoms (e.g., psychosis and anhedonia), executive function and cognitive deficits (e.g., verbal ability and working memory), and developmental processes. Furthermore, using these neuroimaging approaches in conjunction with behavioral phenotypic data in "unsupervised" machine learning models has revealed patterns of neural and behavioral relationships without explicit guidance on what those relationships should be, potentially providing insights into the circuits underlying CNS disorder symptoms (see Ji, et al., Elife 2021, Jul 20:10:e66968).

[0008] U.S. Patent Application Publication 2021 / 0005306 describes the computation of a neurobehavioral geometry consisting of neural features statistically mapped to defined behavioral characteristics in a group of individuals exhibiting a variety of symptoms. This publication describes a method for determining therapeutic drugs for patients with neurological disorders via such neurobehavioral geometry, comprising first deriving unsupervised, data-organized neural maps associated with behavioral dimensions in a group of patients along their CNS disorder spectrum, and then identifying candidate therapeutic targets associated with the derived neural maps. Thus, potential therapeutic drugs for new patients can be identified based on the patient's position along one or more dimensions in the neurobehavioral geometry. However, this publication does not describe a method for a “supervised” approach (i.e., where an algorithm for detecting neurobehavioral relationships is constrained that neural maps must be maximized in similarity to a particular target map). In other words, the publication does not describe how to determine the set of behaviors (or individuals possessing those behaviors) that best align with the neural map for a given target.

[0009] In addition to non-invasive neuroimaging, neural circuits can be characterized by information at the microscale level, such as by using the expression levels of different genes in the brain. In recent work, a spatially comprehensive atlas of the neural transcriptome has been measured using postmortem human brains (see Hawrylycz, et al., 2012, Nature 489:391-99). These transcriptome data are being combined with state-of-the-art imaging methods to leverage planar cortical topography and generate gene expression maps that may relate to brain activity associated with a desired behavioral phenotype (see Burt et al., 2018, Nature Neuroscience 21:1251-59). U.S. Patent No. 11,791,016 describes an approach that can confirm the relationship between neural phenotypic maps representing behavior-related brain activity and gene expression “target” maps associated with drug targets. Specifically, this association between the two maps can be quantified using similarity scores for each phenotype-gene pair, and information on potential therapeutic targets for patients exhibiting a particular symptom or disorder can be obtained by using a gene target map that is highly similar to a neuronal phenotype map for that symptom or disorder. However, the methods described in this published patent require that both the gene expression target map and the patient's neuronal phenotype map be characterized independently beforehand. Therefore, while this can be used to confirm the association between a set of phenotypes and a set of gene expression maps associated with therapeutic targets, it cannot be used to calculate phenotypes associated with a given therapeutic target map. Furthermore, these disclosed methods do not provide a way to derive the relative importance of a particular symptom / behavior scale with respect to a target neuronal target map of interest, such as a gene expression map (i.e., the methods disclosed in the published patent cannot identify, given a particular neuronal target map, which symptoms and / or behaviors are most relevant to the neural circuits most similar to that target map).

[0010] Novel methods and strategies are needed to align sets of behavioral characteristics and associated patterns of their brain activity with specific neural target maps, and to identify subjects for inclusion and / or exclusion in clinical trials based on the determined likelihood of subjects responding to a drug and / or placebo. Novel methods and strategies are also needed to optimize subject enrollment in clinical trials. The present invention relates to further solutions that address these needs and have other desirable features. [Overview of the project]

[0011] A novel method is disclosed herein that facilitates the identification of subjects who are likely to respond to experimental treatment (e.g., a drug) and, as a result, be included in a clinical trial. In certain embodiments, the method disclosed herein may be used to facilitate the identification of subjects who are likely to respond to a placebo and, as a result, be excluded from a clinical trial. In contrast to conventional strategies, the method disclosed herein provides a set of behavioral features that reflect patterns of brain activity, which are then aligned with a designated neural target map. Also in contrast to conventional strategies, the method disclosed herein requires only the evaluation of the behavioral features of candidate subjects to predict their likelihood of responding to experimental treatment. These behavioral features can then be used as registration criteria to enhance clinical trials with subjects exhibiting the behaviors most likely to be associated with neural targets.

[0012] In certain embodiments, the invention disclosed herein relates to a method for selecting subjects as candidate subjects in a clinical trial based on their predicted response to an experimental treatment or therapeutic agent, the method comprising: (a) retrieving a neural monitoring target benchmark map associated with an experimental treatment or therapeutic agent by at least one processor of a computing device, wherein the neural monitoring target benchmark map includes neural target data collected from one or more individuals, the neural target data is associated with an experimental treatment or therapeutic agent, and the neural monitoring target benchmark map further includes a two-dimensional plane and subcortical volume structure, the two-dimensional plane and subcortical volume structure includes one or more numerical values ​​assigned to specific brain locations represented by the two-dimensional plane and subcortical volume structure; and (b) retrieving a reference neural behavior-related dataset by at least one processor of a computing device, wherein the reference neural behavior-related dataset includes neural target data collected from one or more individuals, the neural target data is associated with mental health or cognitive state A step including neuronal features and symptom features related to (or, in a particular case, corresponding to) a neurosurveillance target benchmark map and a reference neurobehavior-related dataset, wherein at least one processor of the computing device determines a symptom-versus-alignment prediction function that reflects the alignment relationship between a neurosurveillance target benchmark map and a reference neurobehavior-related dataset, the determination comprising, via at least one processor of the computing device, calculating a neuronal alignment score of the reference neurobehavior-related dataset relative to the neurosurveillance target benchmark map, wherein a high absolute neuronal alignment score of the reference neurobehavior-related dataset relative to the neurosurveillance target benchmark map indicates a statistical correspondence between neuronal features in the reference neurobehavior dataset and neuronal target data in the neurosurveillance target benchmark map, and the neuronal alignment score is used to calculate a symptom-versus-alignment prediction function that indicates a statistical correspondence between symptom features in the reference neurobehavior dataset and the neuronal alignment score.(d) A step of determining a neural alignment score for a subject using at least one processor of a computing device, the determination comprising evaluating a symptom-versus-alignment prediction function for the subject's symptoms and / or evaluating a neural surveillance target benchmark map for the subject's neural feature map, wherein a high neural alignment score indicates a high probability of alignment with the neural surveillance target benchmark map, and based on this high probability of alignment, the subject is given a quantitative likelihood of response to the experimental treatment or therapeutic agent. It is important to note that the neural surveillance target benchmark map is a priori specified and serves as a constraint for subsequent steps described herein, since the goal of this method is to select a subject having the greatest predicted neural similarity to the neural surveillance target benchmark map (e.g., a subject whose neural feature map is most similar to the selected neural surveillance target benchmark map).

[0013] The method described in the preceding paragraphs and further disclosed herein is readily distinguishable from the “unsupervised” method described in U.S. Patent Application Publication 2021 / 0005306 (whose disclosure is incorporated herein by reference), which prioritizes first defining the relationship between the symptom / behavioral variance dimension and the neural features derived from neural data collected in a population of individuals in order to derive a neurobehavioral map, and then identifying candidate therapeutic targets associated with the derived neurobehavioral map. The supervised method described in this application selects targets that maximize similarity to a pre-specified target benchmark map, whereas the unsupervised method in U.S. Patent Application Publication 2021 / 0005306 selects target benchmark maps based on their similarity to symptom-relevant neurobehavioral maps.

[0014] In certain embodiments of the method described above, a high quantitative likelihood of response indicates that the subject is selected as a candidate subject that is predicted to respond to the experimental treatment or therapeutic agent. Conversely, in certain embodiments of the method described above, a low quantitative likelihood of response indicates that the subject is selected as a candidate subject that is not predicted to respond to the experimental treatment or therapeutic agent.

[0015] In certain embodiments, the method described above may further include a step of selecting candidate subjects who are likely to respond to the experimental treatment or therapeutic agent before randomization in a clinical trial. For example, in certain embodiments, candidate subjects predicted to have the highest neuronal similarity to the neurosurveillance target benchmark map are selected for inclusion in a clinical trial and randomized to receive the experimental treatment (e.g., a drug) or a placebo. Alternatively, in other embodiments, the method described above may further include a step of excluding candidate subjects who are unlikely to respond to the experimental treatment or therapeutic agent before randomization in a clinical trial. For example, in certain embodiments, subjects predicted to have low neuronal similarity to the neurosurveillance target benchmark map are excluded from the clinical trial. In some embodiments, the method disclosed herein may further include a step of administering the experimental treatment or therapeutic agent to subjects selected as candidate subjects predicted to respond to the experimental treatment or therapeutic agent. In some embodiments of the method described above, if a subject is selected as a candidate subject not predicted to respond to the experimental treatment or therapeutic agent, the experimental treatment or therapeutic agent is not administered to the subject.

[0016] In some embodiments of the methods disclosed herein, the neurosurveillance target benchmark map includes a whole-brain PET map of the receptor occupancy of a therapeutic agent. In certain embodiments, the neurosurveillance target benchmark map includes a pharmacological map associated with one or more receptor targets. In other embodiments, the neurosurveillance target benchmark map includes a gene expression map associated with one or more gene expression targets. In other embodiments, the neurosurveillance target benchmark map includes a previously calculated neurobehavioral variability map associated with one or more symptoms and / or signs. In other embodiments, the neurosurveillance target benchmark map includes a task-induced neuromap associated with one or more functions.

[0017] In certain embodiments, the neural surveillance target benchmark map is located in space, where the left and right hemispheres of the target are represented as a plane and the subcortex as a volume. In certain embodiments, the neural surveillance target benchmark map and the reference neural behavior-related dataset are located in the same space. In certain embodiments, the neural surveillance target benchmark map and the reference neural behavior-related dataset are segmented according to an atlas of regions defined by function. Some examples of such regions include the primary visual cortex, the fusiform facial region, the medial prefrontal cortex, and the ring-Sylvian sulcus language region. These regions are defined not only based on neuroanatomical and ultrastructural characteristics but also on functional measures such as functional MRI.

[0018] Alternatively, in other embodiments, the neural monitoring target benchmark map may be at the network, area, or individual vertex or voxel level.

[0019] In certain embodiments, the symptom-versus-alignment prediction function is a linear regression model that generates a set of linear weights to predict neurological alignment scores from a symptom scale. In other embodiments, the symptom-versus-alignment prediction function is a nonlinear mapping that predicts neurological alignment scores from a symptom scale.

[0020] A method for selecting subjects predicted to respond to a placebo is also disclosed herein, the method comprising: (a) a step of extracting a placebo-related neurosurveillance target benchmark map by at least one processor of a computing device, wherein the placebo-related neurosurveillance target benchmark map includes neurotarget data collected from one or more individuals who participated in a clinical trial including a placebo group, the placebo-related neurosurveillance target benchmark map includes a two-dimensional plane and a subcortical volume structure, the two-dimensional plane and subcortical volume structure includes one or more numerical values ​​assigned to specific brain locations represented by the two-dimensional plane and subcortical volume structure; (b) a step of extracting a reference neurobehavioral-related dataset by at least one processor of a computing device, wherein the reference neurobehavioral-related dataset includes neurotarget efficacy data collected from one or more individuals who participated in a clinical trial and were randomized to a placebo group, the neurotarget efficacy data includes neuronal characteristics and symptom characteristics; and (c) a step of extracting a reference neurobehavioral-related dataset by at least one processor of a computing device (d) A process of determining a symptom-versus-alignment predictive function that reflects the alignment relationship between a placebo-related neurosurveillance target benchmark map and a reference neurobehavior-related dataset, wherein the determination includes, via at least one processor of a computing device, calculating a neuronal alignment score of the reference neurobehavior-related dataset relative to the placebo-related neurosurveillance target benchmark map, wherein a high absolute neuronal alignment score of the reference neurobehavior-related dataset relative to the placebo-related neurosurveillance target benchmark map indicates a statistical correspondence between neuronal target efficacy data in the reference neurobehavior-related dataset and neuronal target data in the placebo-related neurosurveillance target benchmark map, and the neuronal alignment score is used to calculate a symptom-versus-alignment predictive function that indicates a statistical correspondence between symptom features in the reference neurobehavior dataset and the neuronal alignment score; and (d) via at least one processor of a computing device,The process includes determining the neural alignment score of the subject, wherein the determination includes evaluating a symptom-versus-alignment predictive function for the subject's symptoms and / or evaluating a placebo-associated neural surveillance target benchmark map for the subject's neural data, wherein a high neural alignment score indicates a high probability of alignment with the placebo-associated neural surveillance target benchmark map, and based on this high probability of alignment, the subject is given a quantitative likelihood of a response to placebo.

[0021] In a particular embodiment, subjects identified with a high neural alignment score were more likely to be placebo responders, while subjects with a low neural alignment score were less likely to be placebo responders.

[0022] In a particular embodiment, the method described above further includes the step of excluding the subject as a likely placebo responder before randomization in a clinical trial.

[0023] In certain embodiments of the present invention, the placebo-related neurosurveillance target benchmark map includes a whole-brain PET map of receptor occupancy associated with one or more receptor targets. Alternatively, in other embodiments, the placebo-related neurosurveillance target benchmark map includes a pharmacological map associated with one or more receptor targets. In other embodiments, the placebo-related neurosurveillance target benchmark map includes a gene expression map associated with one or more gene expression targets. In certain embodiments, the placebo-related neurosurveillance target benchmark map includes a previously calculated neurobehavioral variability map associated with one or more symptoms and / or signs. In other embodiments, the placebo-related neurosurveillance target benchmark map includes a task-induced neuromap associated with one or more functions. In certain embodiments of the described method, the placebo-related neurosurveillance target benchmark map is in space, with the left and right hemispheres of the subject represented as a plane and the subcortex represented as a volume.

[0024] In a particular embodiment, a method for selecting subjects based on a predicted placebo response is used in combination with the above method for selecting subjects based on a predicted response to a therapeutic agent.

[0025] In certain embodiments disclosed herein, the neurosurveillance target benchmark map and the reference neurobehavior-related dataset reside in the same space. In certain embodiments, the placebo-related neurosurveillance target benchmark map and the reference neurobehavior-related dataset are segmented according to an atlas of regions defined by function. Examples of these regions are shown above and include, for example, the primary visual cortex, the fusiform facial region, the medial prefrontal cortex, and the ring-Sylvian sulcus language region.

[0026] In yet another embodiment of the method described above, the placebo-related neural surveillance target benchmark map is located at the network, area, or individual vertex or voxel level.

[0027] In a specific embodiment of the method disclosed in the present specification, the symptom-to-alignment prediction function is a linear regression model that generates a set of linear weights for predicting a neural alignment score from a symptom scale. In another embodiment, the symptom-to-alignment prediction function is a non-linear mapping that predicts a neural alignment score from a symptom scale.

[0028] A method for selecting subjects as candidate subjects based on the subjects' predicted response to an experimental treatment or therapeutic agent is also disclosed herein, the method comprising: (a) a step of extracting a neurosurveillance target benchmark map by at least one processor of a computing device, wherein the neurosurveillance target benchmark map includes neurosurveillance target efficacy data collected from one or more individuals who participated in a clinical trial of the experimental treatment or therapeutic agent, the clinical trial includes a treatment group and a placebo group, the neurosurveillance target benchmark map includes a two-dimensional plane and a subcortical volume structure, and the two-dimensional plane and subcortical volume structure includes one or more numerical values ​​assigned to specific brain locations represented by the two-dimensional plane and subcortical volume structure; and (b) a step of extracting a reference neurobehavioral-related dataset by at least one processor of a computing device, wherein the reference neurobehavioral-related dataset includes baseline clinical data collected from one or more individuals, the baseline clinical data relating to (or, in certain cases, corresponding to) mental health or cognitive state. (c) a step including neural target effectiveness data and symptom characteristics, and a step of determining a symptom-versus-alignment prediction function that reflects the alignment relationship between a neural monitoring target benchmark map and a reference neural behavior-related dataset using at least one processor of a computing device, wherein determining this includes calculating a neural alignment score of the reference neural behavior-related dataset with respect to the neural monitoring target benchmark map via at least one processor of the computing device, wherein a high absolute neural alignment score of the reference neural behavior-related dataset with respect to the neural monitoring target benchmark map indicates a statistical correspondence between neural target effectiveness data in the reference neural behavior-related dataset and neural target effectiveness data in the neural monitoring target benchmark map, and the neural alignment score is used to calculate a symptom-versus-alignment prediction function that indicates a statistical correspondence between symptom characteristics in the reference neural behavior dataset and the neural alignment score.(d) determining, by at least one processor of a computing device, a neural alignment score for said subject, wherein said determining comprises evaluating a symptom-to-alignment prediction function for a symptom of said subject and / or evaluating a neural monitoring target benchmark map against neural data of said subject, a high neural alignment score indicates a high probability of alignment with said neural monitoring target benchmark map, and based on said high probability of alignment, said subject is provided with a quantitative likelihood of response to said experimental treatment or therapeutic agent;

[0029] In an aspect of the method described in the preceding paragraph, a high quantitative likelihood of response indicates that the subject has been selected as a candidate subject predicted to respond to said experimental treatment or therapeutic agent, and a low quantitative likelihood of response indicates that the subject has been selected as a candidate subject not predicted to respond to said experimental treatment or therapeutic agent.

[0030] In certain embodiments, the method further comprises selecting the candidate subject as likely to respond to said experimental treatment or therapeutic agent prior to randomization in a clinical trial. In other embodiments, the method further comprises excluding the candidate subject as unlikely to respond to said experimental treatment or therapeutic agent prior to randomization in a clinical trial. In still another embodiment, the method further comprises administering said experimental treatment or therapeutic agent to the candidate subject predicted to respond to said experimental treatment or therapeutic agent. In some embodiments, if a subject is selected as a candidate subject not predicted to respond to said experimental treatment or therapeutic agent, said experimental treatment or therapeutic agent is not administered to said subject.

[0031] In certain embodiments, the method for selecting a subject based on the subject's predicted response to an experimental treatment or therapeutic agent is used in combination with a method for selecting subjects predicted to respond to a placebo.

[0032] In some embodiments of the methods disclosed herein, the neurosurveillance target benchmark map includes a whole-brain PET map of receptor occupancy of the therapeutic agent. In some embodiments, the neurosurveillance target benchmark map includes a pharmacological map associated with one or more receptor targets. In other embodiments, the neurosurveillance target benchmark map includes a gene expression map associated with one or more gene expression targets. In some embodiments, the neurosurveillance target benchmark map includes a previously calculated neurobehavioral variability map associated with one or more symptoms and / or signs. In some embodiments, the neurosurveillance target benchmark map includes a task-induced neuromap associated with one or more functions. In another embodiment, the neurosurveillance target benchmark map is in space, with the left and right hemispheres of the subject represented as a plane and the subcortex represented as a volume. In certain embodiments, the neurosurveillance target benchmark map and the reference neurobehavioral-associated dataset are in the same space.

[0033] In certain embodiments, the neural surveillance target benchmark map and reference neural behavior-related dataset are segmented according to an atlas of functionally defined regions, as defined above. Examples of such regions may include the primary visual cortex, fusiform facial region, medial prefrontal cortex, and ring-Sylvian sulcus language region. In other embodiments, the neural surveillance target benchmark map is at the network, area, or individual vertex or voxel level.

[0034] In certain embodiments of the methods disclosed herein, the symptom-versus-alignment prediction function is a linear regression model that generates a set of linear weights that predict neurological alignment scores from a symptom scale. In other embodiments, the symptom-versus-alignment prediction function is a nonlinear mapping that predicts neurological alignment scores from a symptom scale.

[0035] The features and associated advantages of the present invention discussed above will be better understood by referring to the subsequent detailed description of the invention. [Brief explanation of the drawing]

[0036] The aforementioned and other purposes, features, and advantages will become apparent from the subsequent description of the specific embodiments of this disclosure illustrated in the accompanying drawings.

[0037] This patent or application document includes at least one drawing made in color. A copy of this patent or patent application publication containing the color drawing will be provided by the Patent and Trademark Office upon request and payment of the necessary fees. [Figure 1] A diagram of the components and processes of one embodiment of the present invention is shown, in which a selection model can be deployed to enhance clinical trials with candidate subjects based on predicted responses. As shown, the method comprises a “model training” step that utilizes existing data and a “model application” step that can be used in new clinical trials to select candidate subjects based on their predicted response to a therapeutic agent, without requiring the collection of new neuronal data. [Figure 2] This section describes the advantages of using the novel methods disclosed herein in clinical trials. As shown, using such methods in the EMBARC clinical trial (Trivedi, et al., J. Psychiatr. Res., 2016, 78:11-23) would have resulted in better statistical power, shorter study time, and lower costs. [Figure 3A] This section illustrates a typical use case of the novel method disclosed herein, aimed at enhancing clinical trials using candidate subjects who are likely to be responders to therapeutic agents. As shown in Figure 3A, a strong placebo response reduces the effect size, and the lack of exclusion criteria for strong placebo responders weakens the success of the clinical trial. [Figure 3B]This section illustrates a typical use case of the novel method disclosed herein, aimed at enhancing clinical trials with candidate subjects likely to be responders to therapeutic agents. In contrast, the novel method disclosed herein, and in particular, the use of this method to identify subjects predicted to have high alignment scores for inclusion in clinical trials, increases the effect size and reduces the number of subjects required for clinical trials, as shown in Figure 3B. [Figure 4] This document presents a block diagram of the process for implementing a computational framework for augmenting clinical trials using candidate subjects who are likely to be responders to therapeutic agents. As shown in the diagram, the model may be trained on an existing dataset. [Figure 5A] This figure illustrates representative analytical results of the novel method disclosed herein, aimed at enhancing clinical trials using candidate subjects likely to be responders to therapeutic agents. The novel method disclosed herein was applied to data from a clinical trial investigating the effects of sertraline on depression (Trivedi, et al., J. Psychiatr. Res., 2016, 78:11-23). ​​A gene expression map of SLC29A4, the gene encoding the target of sertraline, is illustrated. The color scale of the gene expression map reflects the relative levels of expression (Z-score) in each area of ​​the brain, with yellow indicating positive values ​​(i.e., relatively high expression levels of the SLC29A4 gene) and light blue areas indicating negative values ​​(i.e., relatively low expression levels of the SLC29A4 gene). [Figure 5B] This figure illustrates representative analytical results of the novel method disclosed herein, aimed at enhancing clinical trials using candidate subjects likely to be responders to therapeutic agents. The novel method disclosed herein was applied to data from a clinical trial investigating the effects of sertraline on depression (Trivedi, et al., J. Psychiatr. Res., 2016, 78:11-23). ​​The symptom weights obtained by fitting the symptom-versus-alignment prediction function to the clinical trial data are illustrated. Here, the symptom-versus-alignment prediction function is a linear regression represented by the weights shown in Figure 5B. [Figure 5C] This section illustrates representative analytical results of the novel methods disclosed herein, aimed at enhancing clinical trials using candidate subjects likely to be responders to therapeutic agents. The novel methods disclosed herein were applied to data from a clinical trial investigating the effects of sertraline on depression (Trivedi, et al., J. Psychiatr. Res., 2016, 78:11-23). ​​A legend for the symptoms shown is presented in Figure 5B. [Figure 5D] This figure illustrates representative analytical results of the novel method disclosed herein, aimed at enhancing clinical trials using candidate subjects likely to be responders to therapeutic agents. The novel method disclosed herein was applied to data from a clinical trial investigating the effects of sertraline on depression (Trivedi, et al., J. Psychiatr. Res., 2016, 78:11-23). ​​A histogram of alignment scores is shown, highlighting the 20% of subjects with the highest alignment scores (i.e., subjects that could be selected for inclusion in the clinical trial by the claimed method). [Figure 5E] This section illustrates representative analytical results of the novel method disclosed herein, aimed at enhancing clinical trials using candidate subjects likely to be responders to therapeutic agents. The novel method disclosed herein was applied to data from a clinical trial investigating the effects of sertraline on depression (Trivedi, et al., J. Psychiatr. Res., 2016, 78:11-23). ​​The expected effect size is shown as a function of the proportion of subjects with the highest alignment scores who could be selected for inclusion in the clinical trial by the claimed method. [Figure 5F]This figure illustrates representative analytical results of the novel method disclosed herein, aimed at enhancing clinical trials using candidate subjects likely to be responders to therapeutic agents. The novel method disclosed herein was applied to data from a clinical trial investigating the effects of sertraline in depression (Trivedi, et al., J. Psychiatr. Res., 2016, 78:11-23). ​​The required sample volume is shown from the power calculation of a one-sided two-sample t-test assuming 80% power and a 5% false positive rate. The black dashed line indicates the sample volume required to determine the effect size without applying the claimed method (i.e., based on all available subjects, regardless of the subject's symptom and biomarker profile). Therefore, Figure 5F illustrates the reduction in sample volume made possible by the claimed method. [Figure 6A] This section presents representative use cases of the novel methods disclosed herein aimed at excluding individuals likely to be placebo responders. It illustrates how the lack of exclusion criteria for strong placebo responders weakens the success of clinical trials. In particular, the lack of exclusion criteria for strong placebo responders in clinical trials reduces the effect size. [Figure 6B] This section presents a representative use case of the novel method disclosed herein, aimed at excluding individuals likely to be placebo responders. As shown in Figure 6B, the use of the method disclosed herein reduces the observed placebo response in the remaining trial participants, improving the success of the trial. By excluding subjects predicted to have a high placebo response from the clinical trial, the effect size may increase and the number of subjects required for the clinical trial may decrease, as shown in Figure 6B. [Figure 7] This document presents a block diagram of the process for implementing a computational framework for excluding candidate subjects predicted to have a high placebo response from clinical trials. [Figure 8A] This figure illustrates representative analytical results of a novel method disclosed herein, aimed at excluding candidate subjects predicted to have a high placebo response from clinical trials. Symptom weights obtained by fitting a symptom-versus-alignment predictive function to the trial data are shown. [Figure 8B] This section illustrates representative analytical results of the novel method disclosed herein, aimed at excluding candidate subjects predicted to have a high placebo response from clinical trials. A legend for the symptoms shown in Figure 8A is provided. [Figure 8C] A representative analysis of the novel method disclosed herein, aimed at excluding candidate subjects predicted to have a high placebo response from clinical trials, is illustrated. A histogram of alignment scores is shown, with 80% of subjects with the lowest alignment scores selected for exclusion by the claimed method, and the remaining subjects (i.e., those selected for inclusion) highlighted. [Figure 8D] This figure illustrates representative analytical results of the novel method disclosed herein, aimed at excluding candidate subjects predicted to have a high placebo response from clinical trials. The expected effect size is shown as a function of the proportion of candidate subjects with the highest alignment scores that can be selected for exclusion by the claimed method. The effect size is calculated as Cohen's d of the inverse variance weighted mean obtained in the 10-repeat 2-fold cross-validation test set of the model. A negative value indicates a reduction in symptom score, so a lower effect size is better. The black dashed line shows the effect size when all subjects are selected for inclusion (i.e., the claimed method is not applied). [Figure 8E] This figure illustrates representative analytical results of the novel method disclosed herein, aimed at excluding candidate subjects predicted to have a high placebo response from clinical trials. It shows the required sample volume from the power calculation of a one-sided two-sample t-test assuming 80% power and a 5% false positive rate. The black dashed line indicates the sample volume required when determining the effect size without applying the claimed method (i.e., based on all available subjects, regardless of the subject's symptom and biomarker profile). Therefore, Figure 8E illustrates the reduction in required sample volume made possible by the claimed method. [Figure 9A]This document presents representative analytical results of a novel method disclosed herein aimed at increasing the effect size in clinical trials by both reinforcing it in subjects likely to respond and excluding subjects predicted to have a high placebo response. A gene expression map of SLC29A4, the gene encoding the sertraline target, is shown, and this gene expression map was used as the target map for the reinforcement model. [Figure 9B] This document presents representative analytical results of a novel method disclosed herein aimed at increasing effect sizes in clinical trials by both reinforcing them in individuals likely to respond and excluding subjects predicted to have a high placebo response. The reinforcement model, here showing symptom weights obtained by fitting linear regression to the trial data, is presented. [Figure 9C] This paper presents representative analytical results of a novel method disclosed herein that aims to increase effect sizes in clinical trials by both reinforcing them with those likely to respond and excluding subjects predicted to have a high placebo response. The symptom weights are shown in the placebo model, here derived from fitting linear regression to the trial data. [Figure 9D] This document presents representative analytical results of a novel method disclosed herein that aims to increase the effect size in clinical trials by both reinforcing it in subjects likely to respond and excluding subjects predicted to have a high placebo response. A legend of the symptoms shown in Figures 9B and 9C is provided. [Figure 9E] Representative analytical results of the novel method disclosed herein, which aims to increase the effect size in clinical trials by both reinforcing it with subjects likely to respond and excluding subjects predicted to have a high placebo response, are presented. The expected effect size is shown as a function of the proportion of subjects with the highest alignment score who can be selected for inclusion in the reinforcement model and for exclusion in the placebo model, and the black dashed line shows the effect size when all subjects can be selected for inclusion (i.e., the claimed method is not applied). [Figure 9F] Representative analytical results of the novel method disclosed herein, aimed at increasing the effect size in clinical trials by both reinforcing it with those most likely to respond and excluding subjects predicted to have a high placebo response, are presented. The required sample volume from the power calculation of a one-sided two-sample t-test assuming 80% power and a false positive rate of 5% is shown, and the black dashed line indicates the required sample volume when the effect size is determined without applying the claimed method (i.e., based on all available subjects, regardless of symptom and biomarker profiles). As shown in Figure 9F, the claimed method allows for a reduction in sample volume. [Figure 10A] This document presents representative use cases of the novel method disclosed herein, which utilizes response-related target maps. It also illustrates conventional clinical studies that lack an analytical plan for mapping neuropharmaceutical responses and a pathway for improving the results of subsequent trials. [Figure 10B] Representative use cases of the novel method disclosed herein, which utilizes a response relevance target map, are presented. In contrast, as shown in Figure 10B, the method disclosed herein may be used to facilitate the selection of subjects in clinical trials that are more likely to respond to a therapeutic agent. In particular, such a method is used to enhance the precise selection of high-responder candidates. [Figure 11] A block diagram illustrates the process for implementing a computational framework related to selecting targets based on a response relevance target map. [Figure 12A] This section illustrates representative analytical results of the novel methods disclosed herein, aimed at increasing effect size in clinical trials. In particular, a response-related target map was used in this analysis, and the methods disclosed herein were applied to data from a trial investigating the effects of sertraline on depression (Trivedi, et al., J. Psychiatr. Res., 2016, 78:11-23). ​​The symptom-versus-alignment predictive function, here showing symptom weights obtained by fitting a linear regression model to the trial data, is also presented. [Figure 12B] This section illustrates representative analytical results of the novel methods disclosed herein, aimed at increasing effect sizes in clinical trials. In particular, a response-related target map was used in this analysis, and the methods disclosed herein were applied to data from a study investigating the effects of sertraline on depression (Trivedi, et al., J. Psychiatr. Res., 2016, 78:11-23). ​​A legend for the symptoms shown in Figure 12A is provided. [Figure 12C] This figure illustrates representative analytical results of the novel method disclosed herein, aimed at increasing effect size in clinical trials. In particular, a response-related target map was used in this analysis, and the method disclosed herein was applied to data from a trial investigating the effects of sertraline in depression (Trivedi, et al., J. Psychiatr. Res., 2016, 78:11-23). ​​A histogram of alignment scores is shown, highlighting the 30% of subjects with the highest alignment scores (i.e., subjects that could be selected for inclusion by the claimed method). [Figure 12D] This figure illustrates representative analytical results of the novel method disclosed herein, aimed at increasing effect size in clinical trials. In particular, a response-related target map was used in this analysis, and the method disclosed herein was applied to data from a trial investigating the effects of sertraline in depression (Trivedi, et al., J. Psychiatr. Res., 2016, 78:11-23). ​​The expected effect size is shown as a function of the proportion of subjects with the highest alignment scores who can be selected for inclusion by the claimed method. The black dashed line shows the effect size when all subjects are selected for inclusion (i.e., the claimed method is not applied). [Figure 12E]This figure illustrates representative analytical results of the novel method disclosed herein, aimed at increasing effect size in clinical trials. In particular, a response-relevance target map was used in this analysis, and the method disclosed herein was applied to data from a trial investigating the effect of sertraline in depression (Trivedi, et al., J. Psychiatr. Res., 2016, 78:11-23). ​​The required sample volume is shown from the power calculation of a one-sided two-sample t-test assuming 80% power and a 5% false positive rate. The black dashed line indicates the sample volume required to determine the effect size without applying the claimed method (i.e., based on all available subjects, regardless of the symptom and biomarker profile of the subject). As shown in Figure 12E, the claimed method allows for a reduction in sample volume. [Modes for carrying out the invention]

[0038] An illustrative embodiment of the present invention relates to the training and application of a machine learning statistical model that selects candidate targets based on their predicted response to a therapeutic agent. The model is first trained using a reference dataset having neuronal and behavioral data, and target neuronal maps relating to the therapeutic agent, to determine a symptom-versus-alignment prediction function that reflects the alignment between neuro-behavioral data and neuronal target maps in the reference dataset. Specifically, the neuronal alignment score for each target in the reference dataset is calculated using a quantitative measure of spatial similarity between neuronal monitoring target benchmark maps and individual neuronal feature maps, such as Pearson correlation or Spearman rank correlation coefficient. Then, using data from all targets in the reference dataset, a linear and / or nonlinear machine learning statistical model is constructed to predict the target's neuronal alignment score from the target's symptom scale. The performance of the trained model can be evaluated using methods such as independent sample iterations or k-fold cross-validation.

[0039] Next, by applying the symptom-versus-alignment prediction function from the trained model to behavioral data from new candidate subjects, a predicted alignment score using neural target mapping can be calculated, which reflects the probability of the subject's response to the therapeutic agent. Therefore, subjects predicted to have a high probability of response can be screened for clinical trials, increasing the probability of success.

[0040] Figures 1 to 12, where similar parts are designated by the same reference numerals throughout, illustrate exemplary embodiments of the selection of individual subjects for efficacy likelihood in clinical trials through neuroimaging processing and machine learning-based monitoring of clinical scale variability, according to the present invention. While the present invention is described with reference to the exemplary embodiments illustrated in the figures, it should be understood that the present invention can be realized in many alternative forms. Those skilled in the art will understand, in a manner that retains the spirit and scope of the invention, that different methods can be added to modify the parameters of the disclosed embodiments.

[0041] Referring to Figure 1, the present invention includes a statistical model (1) that is initially trained with behavioral data (3) from a reference dataset (2) and alignment scores (5) calculated between neural data from the reference dataset (2) and a neural target map (4) relating to the experimental treatment of interest. The model (1) is then applied to behavioral data (3) from new candidate subjects (6) to calculate predicted alignment scores with neural targets (7). An alignment score threshold (8) is used to determine which candidate subjects to include in the clinical trial and which to exclude from participation in the clinical trial based on their predicted alignment scores.

[0042] During operation, the invention disclosed herein can be used to train models using existing neuroimaging datasets, including existing data from previous clinical trials, and these models can then be applied by clinical trial personnel to screen potential candidates for enrollment in new trials. Once the trained models generate symptom-versus-alignment predictive functions capable of predicting neuronal alignment scores from symptom / behavioral scales, the only data that needs to be collected from new candidates is the corresponding symptom / behavioral data from relevant clinical scales, e.g., mood and depressive symptoms from the Hamilton Depression Rating Scale (HAM-D), psychotic symptoms from the Positive and Negative Syndrome Scale for Schizophrenia (PANSS), or cognitive abilities from the Brief Assessment of Cognition in Schizophrenia (BACS) or Alzheimer's Disease Assessment Scale-Cognitive Subscale (ADAS-Cog).

[0043] The application of the present invention can increase the likelihood of success in clinical trials by enhancing trials with subjects based on their predicted response to experimental treatment and excluding subjects predicted to have a low likelihood of responding to experimental treatment and / or a high likelihood of responding to placebo. In addition, the model that is the subject of the present invention can be trained on existing data and applied to existing clinical trial workflows, thus minimizing the need to interrupt clinical trial protocols. Furthermore, the model to be applied requires only symptom / behavioral data to screen new candidates, and since these symptom / behavioral scales may already be part of the data collection protocol for clinical trials for efficacy and regulatory reasons, the application of the present invention requires little to no additional data collection.

[0044] Figures 5A–5F present representative use cases of applying the novel methods disclosed herein to enhance clinical trials using high-potential responders. As shown in Figures 5A–5F, these methods were applied to data from a trial investigating the effects of sertraline in depression (Trivedi, et al., J. Psychiatr. Res., 2016, 78:11-23). ​​Figure 5A shows the gene expression map of SLC29A4, the gene encoding the target of sertraline. In this specific example, gene expression across the entire brain, including the cerebral hemispheres and subcortical regions, was used, with the exception of the cerebellum due to incomplete data collection for this brain region in the EMBARC dataset. The color scale in the gene expression map reflects the relative level of expression (Z-score) in each area of ​​the brain, with yellow indicating a positive value, i.e., a relatively high expression level of the SLC29A4 gene, and light blue areas indicating a negative value, i.e., a relatively low expression level. This gene expression map was used as a target map for the analysis presented in Figures 5A–5F. Figure 5B shows the symptom weights obtained by fitting the symptom-versus-alignment predictive function to clinical data. The symptom-alignment predictive function is a linear regression represented by the weights shown in Figure 5B. Figure 5C provides a legend for the symptoms shown in Figure 5B. Figure 5D shows a histogram of alignment scores, with the 20% of subjects with the highest alignment scores (i.e., subjects that can be selected for inclusion in a clinical trial by the claimed method) highlighted. The neuronal alignment score is a quantitative measure of spatial similarity between two neuronal maps, here Pearson correlation. Figure 5E shows the expected effect size as a function of the proportion of subjects with the highest alignment scores that can be selected for inclusion in a clinical trial by the claimed method. The effect size is calculated as Cohen's d of the inverse variance weighted mean obtained in a 10-repeat 2-fold cross-validation test set of the model. A negative value indicates a reduction in symptom scores, so a lower effect size is better. The black dashed line indicates the effect size when all subjects are selected for incorporation (i.e., the claimed method is not applied).Figure 5F shows the required sample volume from the power calculation of a one-sided two-sample t-test assuming an 80% power and a 5% false positive rate. The black dashed line shows the sample volume required to determine the effect size without applying the claimed method (i.e., based on all available subjects, regardless of the symptom and biomarker profiles of the subjects). Therefore, this panel illustrates the reduction in sample volume made possible by the claimed method.

[0045] Figures 8A–8E present representative use cases of the novel methods disclosed herein aimed at excluding candidate subjects predicted to have a high placebo response from clinical trials. These methods disclosed herein were applied to data from a trial investigating the effects of sertraline on depression (Trivedi, et al., J. Psychiatr. Res., 2016, 78:11-23). ​​Figure 8A shows the symptom-versus-alignment predictive function, here the symptom weights obtained by fitting a linear regression model to the trial data, and Figure 8B provides a legend for the symptoms shown in Figure 8A. Figure 8C shows a histogram of alignment scores, highlighting that 80% of subjects with the lowest alignment scores were selected for exclusion by the claimed methods, and the remaining subjects, i.e., those selected for inclusion. Figure 8D shows the expected effect size as a function of the proportion of subjects with the highest alignment scores that could be selected for exclusion by the claimed methods. The effect size is calculated as Cohen's d of the inverse variance weighted mean obtained in the test set of 10 replicates 2-fold cross-validation of the model. A negative value indicates a reduction in symptom score, so a lower effect size is better. The black dashed line shows the effect size when all subjects are selected for inclusion (i.e., the claimed method is not applied). Figure 8E shows the required sample volume from the power calculation of a one-sided two-sample t-test assuming 80% power and a 5% false positive rate. The black dashed line shows the required sample volume when determining the effect size without applying the claimed method (i.e., based on all available subjects, regardless of the subject's symptom and biomarker profile). Thus, Figure 8E illustrates the reduction in required sample volume made possible by the claimed method.

[0046] Figures 9A–9F present representative use cases of the novel methods disclosed herein, aimed at reducing the required sample size in clinical trials by both reinforcing those likely to respond and excluding those predicted to have a high placebo response. These methods disclosed herein were applied to data from a trial investigating the effects of sertraline on depression (Trivedi, et al., J. Psychiatr. Res., 2016, 78:11-23). ​​In the current use cases, the reinforcement model (Figure 5) and the placebo-filtered model (Figure 8) were fitted separately to the data, and subjects were selected for inclusion when both the reinforcement model and the placebo-filtered model recommended inclusion. Figure 9A shows the gene expression map of SLC29A4, the gene encoding the target of sertraline. This gene expression map was used as the target map for the reinforcement model. Figure 9B shows the symptom weights obtained from fitting a linear regression to the test data in the reinforcement model, and Figure 9C shows the symptom weights obtained from fitting a linear regression to the test data in the placebo model. Figure 9D provides a legend for the symptoms shown in Figures 9B and 9C. Figure 9E shows the expected effect size as a function of the proportion of subjects with the highest alignment score that are selected for inclusion in the reinforcement model and selected for exclusion in the placebo model. The effect size is calculated as Cohen's d of the inverse variance weighted mean obtained in the test set of 10 repeated 2-fold cross-validation of the model. A negative value indicates a reduction in symptom score, so a lower effect size is better. The black dashed line shows the effect size when all subjects are selected for inclusion (i.e., not applying the claimed method). Figure 9F shows the required sample size from the power calculation of a one-sided 2-sample t-test assuming 80% power and a 5% false positive rate. The black dashed lines indicate the sample volume required to determine the effect size without applying the claimed method (i.e., based on all available subjects, regardless of the symptom and biomarker profile of the subject). Therefore, Figures 9A to 9F illustrate the reduction in sample volume made possible by the claimed method.

[0047] Figures 12A–12E present representative use cases of the novel methods disclosed herein compared to conventional clinical trials. As shown in Figures 12A–12E, these methods were applied to data from a trial investigating the effects of sertraline on depression (Trivedi, et al., J. Psychiatr. Res., 2016, 78:11-23). ​​Figure 12A shows the symptom-versus-alignment predictive function, here the symptom weights obtained by fitting a linear regression model to the trial data, and Figure 12B provides a legend for the symptoms shown in Figure 12A. Figure 12C shows a histogram of alignment scores, with the 30% of subjects with the highest alignment scores (i.e., subjects that can be selected for inclusion by the claimed method) highlighted. Figure 12D shows the expected effect size as a function of the proportion of subjects with the highest alignment scores that can be selected for inclusion by the claimed method. The effect size is calculated as Cohen's d of the inverse variance weighted mean obtained in the test set of 10 repeated 2-fold cross-validation of the model. A negative value indicates a reduction in symptom score, so a lower effect size is better. The black dashed line shows the effect size when all subjects are selected for inclusion, i.e., when the claimed method is not applied. Figure 12E shows the required sample volume from the power calculation of a one-sided two-sample t-test assuming 80% power and a false positive rate of 5%. The black dashed line shows the sample volume required when determining the effect size without applying the claimed method (i.e., based on all available subjects, regardless of the subject's symptom and biomarker profile). Thus, Figure 12E illustrates the reduction in sample volume made possible by the claimed method.

[0048] definition As used herein, the terms “drug” and “therapeutic agent” are used synonymously to refer to any substance that provides medical and / or health benefits, including the prevention and / or treatment of disease, but is not food or part of food.

[0049] As used herein, the term “neuronal surveillance target benchmark map” means a representation of the brain to which values ​​are assigned to specific locations / regions that reflect the molecular, circuit, mechanistic, and / or biological characteristics of the location, relating to or considered relating to the experimental treatment or therapeutic agent being studied, and used to guide the training of machine learning statistical models.

[0050] As used herein, the term “neurobehavioral dataset” means a dataset containing one or more individuals with neurological and symptom / behavioral data scales related to mental health or cognitive status.

[0051] As used herein, the term “neuronal target” means a receptor, circuit, or biological mechanism that is directly or indirectly affected by a therapeutic agent. Examples of neural targets include serotonin receptors, SLC29A / ENT transporter proteins, and presynaptic neurons.

[0052] As used herein, the term “neural alignment score” means a quantitative measure of spatial similarity between two neural maps, for example, a neural surveillance target benchmark map and a neural feature map from a reference neural behavior-related dataset, such as Pearson correlation or Spearman rank correlation coefficient.

[0053] As used herein, the term “neuronal target data” means measured values ​​from one or more modalities in one or more individuals relating to the structure and / or function of the brain, used to relate to a neural surveillance target benchmark map during the training of a machine learning statistical model. Examples include, but are not limited to, functional neural connectivity derived from blood oxygen level-dependent functional magnetic resonance imaging (BOLD MRI) and structural measures such as stochastic tractography from diffusion-weighted MRI.

[0054] As used herein, the term “neuronal target efficacy data” means the neural target data as defined above, collected from participants in a clinical trial and reflecting the effect of the treatment (or placebo) received by those participants.

[0055] As used herein, the term “neuronal feature map” means a representation of the brain of a single subject, in which values ​​at specific locations / regions reflect measures of the molecular, circuit, mechanistic, and / or biological characteristics of the location, or scores derived from statistical or mathematical processes calculated from such measures.

[0056] As used herein, the term “symptom scale” means a measure that quantifies the severity of an individual’s behaviors and / or experiences relating to a mental or neurological disorder or cognitive state. Examples include anesthesia, hallucinations, and working memory deficits.

[0057] As used herein, the term “symptom-versus-alignment predictive function” means a linear or nonlinear computation that codes for a relationship between a symptom / behavioral scale and a neural alignment score, and can therefore be used to predict one from the other. An exemplary linear computation is linear regression, and exemplary nonlinear computations include, but are not limited to, kernel regression and neural networks.

[0058] It should be understood that embodiments of the present invention may be implemented in hardware, firmware, software, or a combination thereof. In such embodiments, various components and processes may be implemented in hardware, firmware, and / or software to perform the functions of the present invention. That is, the same portion of a hardware, firmware, or software module may perform one or more of the illustrated processes or components in the methods disclosed herein. The present invention may be implemented in one or more computer systems capable of performing the functions described herein.

[0059] The computer systems disclosed herein may have or include one or more processors. The processors may be dedicated digital signal processors or general-purpose digital signal processors. In certain embodiments, the processors are connected to a communication infrastructure, such as a bus or network. Various software embodiments are described in relation to this exemplary computer system. Reading this description will make it clear to those skilled in the art how the disclosure may be implemented using other computer systems and / or computer architectures.

[0060] A computer system also includes main memory, preferably random-access memory (RAM), and may also include secondary memory. The secondary memory may include, for example, a hard disk drive and / or a removable storage drive, such as a floppy disk drive, magnetic tape drive, optical disk drive, or solid-state disk. The removable storage drive reads from and / or writes to the removable storage unit in a well-known manner. The removable storage unit may be a floppy disk, magnetic tape, optical disk, or solid-state disk, which is read from and written to by the removable storage drive. As will be understood by those skilled in the art, the removable storage unit may include a computer-usable storage medium containing computer software and / or data.

[0061] In alternative embodiments, the secondary memory may include other similar means for enabling the loading of computer programs or other instructions into the computer system. Such means may include, for example, removable storage units and interfaces. Examples of such means may include program cartridges and cartridge interfaces (such as those found in video game devices), removable memory chips and associated sockets, thumb drives and USB ports, and other removable storage units and interfaces that enable the transfer of software and data from the removable storage unit to the computer system.

[0062] This computer system may also include a communication interface. This communication interface allows software and data to be transferred between the computer system and external devices. Examples of communication interfaces may include modems, network interfaces (such as Ethernet cards), communication ports, PCMCIA slots and cards, etc. The software and data transferred via the communication interface are in the form of signals, which may be electronic signals, electromagnetic signals, optical signals, or other signals that can be received by the communication interface. These signals are provided to the communication interface via a communication path. The communication path carries the signals and may be implemented using wires or cables, optical fibers, telephone lines, cell phone links, RF links, and other communication channels.

[0063] As used herein, the terms “computer program medium” and “computer-readable medium” are generally used to refer to tangible storage media such as hard disks installed on removable storage units or hard disk drives. These computer program products are means for providing software to a computer system. The computer program (also called computer control logic) is stored in main memory and / or secondary memory. The computer program may also be received via a communication interface. Execution of such a computer program enables the computer system to implement the methods of the Disclosure as considered herein. In particular, execution of the computer program enables the processor to implement the processes of the Invention, such as any of the novel methods described herein. Thus, such a computer program is a controller of the computer system. If the Disclosure is implemented using software, the software may be stored in a computer program product and loaded into a computer system using a removable storage drive, interface, or communication interface.

[0064] In other embodiments, the features of the Disclosure are implemented primarily in hardware, using hardware components such as application-specific integrated circuits (ASICs) and gate arrays. The provision of a hardware state machine for performing the functions described herein will also be apparent to those skilled in the art.

[0065] Numerous modifications and alternative embodiments of the present invention will become apparent to those skilled in the art in consideration of the foregoing description. Therefore, this description should be interpreted only as illustrative and is intended to teach those skilled in the art the best mode for carrying out the invention. Structural details may vary considerably without departing from the spirit of the invention, and the exclusive use of all modifications that fall within the scope of the appended claims is reserved. While embodiments have been described herein in a manner that enables the writing of a clear and concise specification, it is intended and understood that embodiments may be combined or separated in various ways without departing from the invention. The present invention is intended to be limited only to the extent required by the scope of the appended claims and the rules of applicable law. The following specific examples should be interpreted solely as illustrative and should not be interpreted as limiting the following disclosure in any way. [Examples]

[0066] Example 1: Enhancing clinical trials using symptom scales Figures 5A–5F illustrate representative use cases of applying the novel method disclosed herein to enhance clinical trials of a specified neurosurveillance target benchmark map using highly likely responders. In this example, the method was applied to data from a trial investigating the effects of sertraline in depression (Trivedi, et al., J. Psychiatr. Res., 2016, 78:11-23) and to the gene expression map of SLC29A4, the gene encoding the target of sertraline. Using this gene expression map as the neurosurveillance target benchmark map, a neuronal alignment score was calculated by calculating the similarity of the global brain connectivity (GBC) map of each subject's neuronal features to the SLC29A4 gene expression map. Subsequently, these neuronal alignment scores and the subject's symptom scale were used to train a model capable of predicting a patient's neuronal alignment to the SLC29A4 gene expression map based solely on symptoms. Here, the symptom alignment prediction function is a linear regression represented by weights for each symptom scale. Next, a symptom alignment prediction function is applied to calculate the predicted neurological alignment score from the symptom scale, and the number of subjects with the highest alignment score (e.g., in the top 20% or meeting a pre-specified threshold) is selected for inclusion in the claimed clinical trial. The more selective the inclusion criteria (i.e., the higher the alignment score threshold for subjects that can be selected for inclusion in the claimed clinical trial), the larger the expected effect size and the smaller the sample size required to give power to the clinical trial.

[0067] Example 2: Excluding placebo responders from the clinical trial. Figures 8A–8E illustrate representative use cases of the novel method disclosed herein, aimed at excluding candidate subjects predicted to have a high placebo response from clinical trials. The method disclosed herein was applied to data from a trial investigating the effects of sertraline on depression (Trivedi, et al., J. Psychiatr. Res., 2016, 78:11-23). ​​In this example, the neurosurveillance target benchmark map represents the placebo response to serotonin, and subjects with the lowest alignment scores are selected for exclusion using the claimed method (i.e., because these subjects have a higher likelihood of being placebo responders). The remaining subjects may be included in the clinical trial.

[0068] Example 3: Reduce the sample size required to enhance the clinical trial by both enhancing high-potential responders and excluding placebo responders. Figures 9A–9F present representative use cases of the novel method disclosed herein, which aims to reduce the required sample size in clinical trials by both reinforcing those likely to respond and excluding those predicted to have a high placebo response. The method disclosed herein was applied to data from a trial investigating the effects of sertraline in depression (Trivedi, et al., J. Psychiatr. Res., 2016, 78:11-23). ​​In this example, both the reinforcement model (Example 1) and the placebo-filtered model (Example 2) were fitted separately to the data, and subjects were selected for inclusion if both the reinforcement model and the placebo-filtered model recommended inclusion.

[0069] Example 4: Enhancing a new clinical trial using results from a previous clinical trial. Figures 12A–12E present representative use cases of the novel method disclosed herein compared to conventional clinical trials. As shown in Figures 12A–12E, the method was applied to data from a trial investigating the effects of sertraline in depression (Trivedi, et al., J. Psychiatr. Res., 2016, 78:11-23). ​​In this example, the neurosurveillance target benchmark map is a map of neuronal responses to experimental treatment from a previously conducted trial. By using this method, new clinical trials can leverage information from previous trials to select patients enhanced for high predicted neuronal alignment against the response map of experimental treatment. This increases the expected effect size and reduces the number of subjects required to give power to the clinical trial.

[0070] It should be understood that the following claims are intended to encompass all the general and specific features of the invention described herein, as well as all descriptions of the scope of the invention that may be considered to fall under those categories as a matter of language.

Claims

1. A method for selecting a subject as a candidate subject based on the subject's predicted response to an experimental treatment or therapeutic agent, a) Extracting a neural surveillance target benchmark map associated with the experimental treatment or therapeutic agent using at least one processor of a computing device, i) The neural surveillance target benchmark map includes neural target data collected from one or more individuals, and the neural target data is associated with the experimental treatment or therapeutic agent. ii) The neural surveillance target benchmark map further includes a two-dimensional plane and a subcortical volume structure, wherein the two-dimensional plane and the subcortical volume structure include one or more numerical values ​​assigned to specific brain locations represented by the two-dimensional plane and the subcortical volume structure, b) Extracting a reference neurobehavior-related dataset using at least one processor of the computing device, i) The standard neurobehavioral dataset includes neural target data collected from one or more individuals, and the neural target data includes neural characteristics and symptom characteristics related to mental health or cognitive state. c) Determining a symptom-versus-alignment prediction function that reflects the alignment relationship between the neurosurveillance target benchmark map and the reference neurobehavior-related dataset, using at least one processor of the computing device, i) The determination includes calculating the neural alignment score of the reference neural behavior-related dataset for the neural monitoring target benchmark map via the at least one processor of the computing device, ii) A high absolute neural alignment score between the reference neural behavior-related dataset and the neural monitoring target benchmark map indicates a statistical correspondence between the neural features in the reference neural behavior dataset and the neural target data in the neural monitoring target benchmark map. iii) The neural alignment score is used to calculate the symptom-versus-alignment prediction function, which shows a statistical correspondence between symptom features in the reference neurobehavioral dataset and the neural alignment score. d) Determining the neural alignment score of the target using at least one processor of the computing device, i) The determination includes evaluating the symptom-versus-alignment prediction function for the target symptom and / or evaluating the neurosurveillance target benchmark map for the target neuronal feature map, ii) A high neural alignment score indicates a high probability of alignment with the neural monitoring target benchmark map. iii) A method comprising: providing the subject with a quantitative likelihood of response to the experimental treatment or therapeutic agent based on the high probability of the alignment.

2. The method according to claim 1, wherein a high quantitative likelihood of response indicates that the subject is selected as a candidate subject that is predicted to respond to the experimental treatment or therapeutic agent.

3. The method according to claim 1, wherein a low quantitative likelihood of response indicates that the subject is selected as a candidate subject not expected to respond to the experimental treatment or therapeutic agent.

4. The method according to claim 2, further comprising the step of selecting candidate subjects who are likely to respond to the experimental treatment or therapeutic agent before randomization in a clinical trial.

5. The method according to claim 3, further comprising the step of excluding candidate subjects who are unlikely to respond to the experimental treatment or therapeutic agent before randomization in a clinical trial.

6. The method according to claim 2, further comprising the step of administering the experimental treatment or therapeutic agent to a candidate subject selected as a candidate subject that is expected to respond to the experimental treatment or therapeutic agent.

7. The method according to claim 3, wherein if the subject is selected as a candidate subject that is not expected to respond to the experimental treatment or therapeutic agent, the experimental treatment or therapeutic agent is not administered to the subject.

8. The method according to claim 1, wherein the neural surveillance target benchmark map includes a whole-brain PET map of receptor occupancy of the therapeutic agent.

9. The method according to claim 1, wherein the neurosurveillance target benchmark map includes a pharmacological map associated with one or more receptor targets.

10. The method according to claim 1, wherein the neural surveillance target benchmark map includes a gene expression map associated with one or more gene expression targets.

11. The method according to claim 1, wherein the neurosurveillance target benchmark map includes a previously calculated neurobehavioral variability map associated with one or more symptoms and / or signs.

12. The method according to claim 1, wherein the neural monitoring target benchmark map includes task-induced neural maps associated with one or more functions.

13. The method according to claim 1, wherein the neural surveillance target benchmark map is in space, with the target left and right hemispheres represented as a plane and the subcortex represented as a volume.

14. The method according to claim 1, wherein the neural monitoring target benchmark map and the reference neural behavior-related dataset are in the same space.

15. The method according to claim 1, wherein the neural monitoring target benchmark map and the reference neural behavior-related dataset are divided according to an atlas of regions defined by function.

16. The method according to claim 1, wherein the neural monitoring target benchmark map is at the network, area, or individual vertex or voxel level.

17. The method according to claim 1, wherein the symptom-versus-alignment prediction function is a linear regression model that generates a set of linear weights to predict neurological alignment scores from a symptom scale.

18. The method according to claim 1, wherein the symptom-to-alignment prediction function is a nonlinear mapping that predicts a neurological alignment score from a symptom scale.

19. A method for selecting subjects that are predicted to respond to a placebo, a) Extracting a placebo-related neural surveillance target benchmark map using at least one processor of a computing device, i) The placebo-related neurosurveillance target benchmark map includes neurotarget data collected from one or more individuals who participated in a clinical trial including the placebo group, ii) The placebo-associated neural surveillance target benchmark map further comprises a two-dimensional plane and a subcortical volume structure, wherein the two-dimensional plane and the subcortical volume structure include one or more numerical values ​​assigned to specific brain locations represented by the two-dimensional plane and the subcortical volume structure, b) Extracting a reference neurobehavior-related dataset using at least one processor of the computing device, i) The reference neurobehavioral dataset includes neuronal target efficacy data collected from one or more individuals who participated in the clinical trial and were randomized to the placebo group, and the neuronal target efficacy data includes neuronal characteristics and symptom characteristics. c) Determining a symptom-versus-alignment predictive function that reflects the alignment relationship between the placebo-associated neurosurveillance target benchmark map and the reference neurobehavior-related dataset, using at least one processor of the computing device, i) The determination includes, via the at least one processor of the computing device, calculating the neural alignment score of the reference neural behavior-related dataset against the placebo-related neural monitoring target benchmark map, ii) The high absolute neural alignment score of the reference neural behavior-related dataset compared to the placebo-related neural surveillance target benchmark map indicates a statistical correspondence between the neural target efficacy data in the reference neural behavior-related dataset and the neural target data in the placebo-related neural surveillance target benchmark map. iii) The neural alignment score is used to calculate the symptom-versus-alignment prediction function, which shows a statistical correspondence between symptom features in the reference neurobehavioral dataset and the neural alignment score. d) Determining the neural alignment score of the target using at least one processor of the computing device, i) The determination includes evaluating the symptom versus alignment prediction function for the subject symptom and / or evaluating the placebo-related neurosurveillance target benchmark map for the subject neuronal data, ii) A high neural alignment score indicates a high probability of alignment with the placebo-associated neural surveillance target benchmark map. iii) A method comprising: providing the subject with a quantitative likelihood of a response to a placebo based on the high probability of the alignment.

20. The method according to claim 19, wherein subjects with a high neural alignment score are more likely to be placebo responders.

21. The method according to claim 19, wherein subjects with a low neural alignment score are less likely to be placebo responders.

22. The method according to claim 20, further comprising the step of excluding the subject as a likely placebo responder before randomization in a clinical trial.

23. The method of claim 19, further comprising selecting the subject as a candidate subject according to the method of claim 1.

24. The method according to claim 19, wherein the placebo-associated neurosurveillance target benchmark map includes a whole-brain PET map of receptor occupancy associated with one or more receptor targets.

25. The method according to claim 19, wherein the placebo-related neurosurveillance target benchmark map includes a pharmacological map associated with one or more receptor targets.

26. The method according to claim 19, wherein the placebo-related neural surveillance target benchmark map includes a gene expression map associated with one or more gene expression targets.

27. The method according to claim 19, wherein the placebo-associated neurosurveillance target benchmark map includes a previously calculated neurobehavioral variability map associated with one or more symptoms and / or signs.

28. The method according to claim 19, wherein the placebo-associated neural surveillance target benchmark map includes task-induced neural maps associated with one or more functions.

29. The method according to claim 19, wherein the placebo-related neural surveillance target benchmark map is in space, with the left and right hemispheres of the target represented as a plane and the subcortex represented as a volume.

30. The method according to claim 19, wherein the placebo-related neurosurveillance target benchmark map and the reference neurobehavior-related dataset are in the same space.

31. The method according to claim 19, wherein the placebo-related neurosurveillance target benchmark map and the reference neurobehavior-related dataset are divided according to an atlas of regions defined by function.

32. The method according to claim 19, wherein the placebo-related neural surveillance target benchmark map is at the network, area, or individual vertex or voxel level.

33. The method according to claim 19, wherein the symptom-versus-alignment prediction function is a linear regression model that generates a set of linear weights to predict neurological alignment scores from a symptom scale.

34. The method according to claim 19, wherein the symptom-to-alignment prediction function is a nonlinear mapping that predicts a neurological alignment score from a symptom scale.

35. A method for selecting a subject as a candidate subject based on the subject's predicted response to an experimental treatment or therapeutic agent, a) Extracting a neural monitoring target benchmark map using at least one processor of a computing device, i) The neurosurveillance target benchmark map includes neurosurveillance target efficacy data collected from one or more individuals who participated in the experimental treatment or clinical trial of the therapeutic agent, and the clinical trial includes a treatment group and a placebo group, ii) The neural surveillance target benchmark map further includes a two-dimensional plane and a subcortical volume structure, wherein the two-dimensional plane and the subcortical volume structure include one or more numerical values ​​assigned to specific brain locations represented by the two-dimensional plane and the subcortical volume structure, b) Extracting a reference neurobehavior-related dataset using at least one processor of the computing device, i) The standard neurobehavioral dataset includes baseline clinical data collected from one or more individuals, and the baseline clinical data includes neuronal target efficacy data and symptom characteristics related to mental health or cognitive status. c) Determining a symptom-versus-alignment prediction function that reflects the alignment relationship between the neurosurveillance target benchmark map and the reference neurobehavior-related dataset, using at least one processor of the computing device, i) The determination includes calculating the neural alignment score of the reference neural behavior-related dataset for the neural monitoring target benchmark map via the at least one processor of the computing device, ii) A high absolute neural alignment score between the reference neural behavior-related dataset and the neural monitoring target benchmark map indicates a statistical correspondence between the neural target effectiveness data in the reference neural behavior-related dataset and the neural target effectiveness data in the neural monitoring target benchmark map. iii) The neural alignment score is used to calculate the symptom-versus-alignment prediction function, which shows a statistical correspondence between symptom features in the reference neurobehavioral dataset and the neural alignment score. d) Determining the neural alignment score of the target using at least one processor of the computing device, i) The determination includes evaluating the symptom-versus-alignment prediction function for the target symptom and / or evaluating the neurosurveillance target benchmark map for the target neuronal data, ii) A high neural alignment score indicates a high probability of alignment with the neural monitoring target benchmark map. iii) A method comprising: providing the subject with a quantitative likelihood of response to the experimental treatment or therapeutic agent based on the high probability of the alignment.

36. The method according to claim 35, wherein a high quantitative likelihood of response indicates that the subject is selected as a candidate subject that is predicted to respond to the experimental treatment or therapeutic agent.

37. The method according to claim 35, wherein a low quantitative likelihood of response indicates that the subject is selected as a candidate subject not expected to respond to the experimental treatment or therapeutic agent.

38. The method according to claim 36, further comprising the step of selecting candidate subjects who are likely to respond to the experimental treatment or therapeutic agent before randomization in a clinical trial.

39. The method according to claim 37, further comprising the step of excluding candidate subjects who are unlikely to respond to the experimental treatment or therapeutic agent before randomization in a clinical trial.

40. The method according to claim 36, further comprising the step of administering the experimental treatment or therapeutic agent to the candidate subject.

41. The method according to claim 37, wherein if the subject is selected as a candidate subject that is not expected to respond to the experimental treatment or therapeutic agent, the experimental treatment or therapeutic agent is not administered to the subject.

42. The method of claim 35, further comprising selecting subjects who are expected to respond to a placebo according to the method of claim 19.

43. The method according to claim 35, wherein the neural surveillance target benchmark map includes a whole-brain PET map of receptor occupancy of the therapeutic agent.

44. The method according to claim 35, wherein the neurosurveillance target benchmark map includes a pharmacological map associated with one or more receptor targets.

45. The method according to claim 35, wherein the neural surveillance target benchmark map includes a gene expression map associated with one or more gene expression targets.

46. The method according to claim 35, wherein the neurosurveillance target benchmark map includes a previously calculated neurobehavioral variability map associated with one or more symptoms and / or signs.

47. The method according to claim 35, wherein the neural monitoring target benchmark map includes task-induced neural maps associated with one or more functions.

48. The method according to claim 35, wherein the neural surveillance target benchmark map is in space, with the target left and right hemispheres represented as a plane and the subcortex represented as a volume.

49. The method according to claim 35, wherein the neural monitoring target benchmark map and the reference neural behavior-related dataset are in the same space.

50. The method according to claim 35, wherein the neural surveillance target benchmark map and the reference neural behavior-related dataset are divided according to an atlas of regions defined by function.

51. The method according to claim 35, wherein the neural monitoring target benchmark map is at the network, area, or individual vertex or voxel level.

52. The method according to claim 35, wherein the symptom-versus-alignment prediction function is a linear regression model that generates a set of linear weights to predict neurological alignment scores from a symptom scale.

53. The method according to claim 35, wherein the symptom-to-alignment prediction function is a nonlinear mapping that predicts a neurological alignment score from a symptom scale.

Images