Introduction to Quantitative EEG and Neurofeedback

By James R. Evans

Introduction to Quantitative EEG and Neurofeedback, Third Edition offers a window into brain physiology and function via computer and statistical analyses, suggesting innovative approaches to the improvement of attention, anxiety, mood and behavior. Resources for understanding what QEEG and neurofeedback are, how they are used, and to what disorders and patients they can be applied are scarce, hence this volume serves as an ideal tool for clinical researchers and practicing clinicians. Sections cover advancements (including Microcurrent Electrical Stimulation, photobiomodulation), new applications (e.g. Asperger's, music therapy, LORETA, etc.), and combinations of prior approaches.

New chapters on smart-phone technologies and mindfulness highlight their clinical relevance. Written by top scholars in the field, this book offers both the breadth needed for an introductory scholar and the depth desired by a clinical professional.

• Covers neurofeedback use in depression, ADHD, addiction, pain, PTSD, and more
• Discusses the use of adjunct modalities in neurotherapy
• Features topics relevant to the knowledge blueprints for both the International QEEG Certification Board and International Board of Quantitative Electrophysiology
• Includes new chapters on photobiomodulation, smart-phone applications and mindfulness

• Language: English
• Publisher: Academic Press
• Release date: Jun 27, 2023
• ISBN: 9780323984331

Book preview

Section I

Raw EEG and QEEG and NFB foundational concepts

Outline

Chapter 1 Electroencephalography microstates in relation to emotional decision-making

Chapter 2 Quantitative electroencephalography (qEEG) clinical applications: a review and update

Chapter 3 Electroencephalographic imaging and biofeedback training using Z-scores: databases and LORETA-based methods

Chapter 4 Autonomic nervous system and the triple network: an evolutionary perspective with clinical implications

Chapter 5 Taking a deeper look into the wiring of baby humans

Chapter 6 When can you trust beta: detecting electromyography contamination of the electroencephalogram signal to enhance assessment of quantitative electroencephalogram and electroencephalogram training

Chapter 7 Origins of electroencephalogram rhythms and implications for neurofeedback

Chapter 8 History of the scientific standards of quantitative electroencephalography normative databases

Chapter 9 Electroencephalogram neuroimaging, brain networks, and neurofeedback protocols

Chapter 10 Electroencephalography in depth: seeing psyche in brainwaves

Chapter 11 Theories of brain functioning and the quantitative electroencephalography

Chapter 1

Electroencephalography microstates in relation to emotional decision-making

Ronald Bonnstetter¹, Thomas F. Collura², Carlos Zalaquett³ and Huai-Hsuan Wang³,    ¹TTISI Target Training International Success Insights, Scottsdale, AZ, United States,    ²BrainMaster Technologies, Inc., Bedford, OH, United States,    ³The Pennsylvania State University, State College, PA, United States

Abstract

People make daily decisions about purchasing their favorite food, responding to a phone call, or loving someone. Scientists have been studying human’s decision-making patterns for decades. Most of the decision-making research has relied on self-report questionnaires or scales to quantify people’s preferences. Despite the prevalent use of self-report measurements, the correspondence between self-evaluations of choice and actual performance is debatable. Metaanalyses across diverse disciplines and abilities show a moderate mean correlation (M = 0.29) between self-evaluations and performance. Inaccuracy or imprecision of self-reports are also affected by factors such as social desirability, inherent to this type of inquiry.

Keywords

Electroencephalography; positron emission tomography; magnetic resonance imaging; emotional decision-making

I Supporting literature

People make daily decisions about purchasing their favorite food, responding to a phone call, or loving someone. Scientists have been studying human’s decision-making patterns for decades. Most of the decision-making research has relied on self-report questionnaires or scales to quantify people’s preferences. Despite the prevalent use of self-report measurements, the correspondence between self-evaluations of choice and actual performance is debatable. Metaanalyses across diverse disciplines and abilities show a moderate mean correlation (M = 0.29) between self-evaluations and performance (Matsuyoshi & Watanabe, 2021). Inaccuracy or imprecision of self-reports are also affected by factors such as social desirability, inherent to this type of inquiry (Choi & Pak, 2005).

Brain research has provided additional ways to advance our understanding of underlying mechanisms and processes of human behavior, cognition, and emotion (Collura et al., 2014; Piwowarski et al., 2020). Neuroscientific technologies such as functional magnetic resonance imaging, positron emission tomography, electroencephalography (EEG), and neurofeedback helped study the biological substrates of behavior (Collura et al., 2014). Among these technologies, EEG is one of the most effective ways to detect brainwaves. EEG can track high-speed brain activity and detect frontal hemispheric asymmetry, providing a platform to study cognition, motivation, emotions, behavior, and personality (Piwowarski et al., 2020). EEG provides the opportunity to observe brain activity and identify subjective experiences (e.g., emotions, decision-making) by measuring patterns of brainwaves in different parts of the brain.

To understand cognitive control, motivation, emotions, and decision-making, the study of frontal asymmetry became the focus of research (Piwowarski et al., 2020). This focus was based on pioneer studies of emotions, approach-avoidance, and decision-making using EEG performed by Davidson, Tomarken, and colleagues (e.g., Davidson, 1992; Tomarken et al., 1990; Wheeler et al., 1993) and Baehr and colleagues (Baehr, Rosenfeld, & Baehr, 1997; Baehr, Rosenfeld, Baehr, & Earnest, 1997) who showed that frontal activation asymmetry correlates with affect. The work of Davidson, Rosenfeld, and Baehr set the stage for the field frontal asymmetry studies of decision-making (Collura et al., 2014).

The term frontal asymmetry refers to the atypical brain activities that are detected in the frontal cortex between the left and right hemispheres. The study of frontal asymmetry had two approaches. The first involves the study of frontal alpha asymmetry and the variable of resting, which relates to psychological constructs and predicting future emotional actions. The second is studying how the state of changes in frontal asymmetry relates to the current emotional state or behavior (Piwowarski et al., 2020).

The EEG study of frontal asymmetry revealed that the left and right hemispheres are related to different motivations (Coan & Allen, 2003; Harmon-Jones & Allen, 1997; Sutton & Davidson, 1997; Kelley et al., 2017). The left hemisphere facilitates approach behavior, positive affect that is approach-related (e.g., enthusiasm, pride, joy), and moving toward a desired goal. The right hemisphere facilitates the withdrawal behavior from sources of aversive stimulation, certain forms of negative affect that are withdrawal-related (e.g., fear, disgust), and increasing the distance between the person and the source of aversive stimulation (Coan & Allen, 2003; Collura et al., 2014; Dawson et al., 1992; Harmon-Jones & Allen, 1997; Kelley et al., 2017; Sutton & Davidson, 1997).

Lateralized brain-based approach-avoidance behaviors show an evolutionary progression. Simple organisms use these behaviors for survival, whereas more complex organisms use these behaviors to approach reward and avert punishment when making decisions. (Van Honk & Schutter, 2005, 2006). Furthermore, recent findings suggest that cortical activations for approach and avoidance motivation are antagonists and that cerebral lateralization serves to inhibit approach and avoidance behaviors from occurring simultaneously (Kelley et al., 2017).

From a neuroscientific and mental health point of view, emotions are essential in decision-making behaviors. Current decision-making models, and state- and trait-based models, both highlight the impact of emotions. The individual's way of interacting with the internal world and external world can be determined by emotions. In addition, emotions can exceed reasoning and rational thinking in decision-making (Collura et al., 2014). Another model, Haidt’s social intuitive model indicates that intuition, which may be associated with emotions, plays a vital role in the decision-making process. This model focuses on the flow between 2 forms of information, intuition, and judgment, which indicates a framework of person A’s response to the situation, but also person A’s interaction with person B (Collura et al., 2014). Based on the insight in neuroscience, these models could provide new perspectives in evidence-based counseling theories and treatments.

Collura et al. (2014) developed an integrative model to bridge mind and body. Their model is built on the observation that the talk cure changes the brain and mind (Ivey, Ivey, & Zalaquett, 2022; Kandel, 2013). The model represents a mechanism for exploring decisions while the person is directly engaged in choice making, exposing precognitive emotional responses to identified beliefs, thoughts, feelings, and actions. Collura et al. (2014) measured frontal EEG gamma band activity at the precognitive level to describe and evaluate approach-avoidance decision-making (Collura et al., 2014; Collura & Bonnstetter, 2018). Gamma is chosen because it provides an immediate emotional response to a stimulus. This approach evaluates the intensity of a person's emotional response to a stimulus by measuring voxel activation and also provides emotional directionality, even before a conscious thought has formed. Their approach further differentiates approach/withdrawal responses within the prefrontal cortex.

II Initial application

The original application of this protocol was to measure response processing as participants responded to self-report surveys. In 2015 and again in 2016, patents were issued to Bonnstetter et al. for this Validation Processes for Ipsative Assessments process. The patent abstracts read in part:

This invention is a validation process for ipsative assessments. Respondents are connected to an Electroencephalograph (EEG) and some or all of the ipsative assessment questions are asked again while connected to the EEG. The EEG measuring frontal lobe responses in terms of gamma waves is compared with the assessment questions. Positive responses proved one frontal lobe response in terms of gamma waves, negative or false answers provide a different gamma response, and neutral questions provide a neutral gamma response. Reading the responses then tells whether the respondent initially responded with integrity. If so, the assessment is response validated Bonnstetter et al. (2015).

Detailing all protocols used for our internal response validation process is beyond the scope of this chapter. The interested reader may consult; Bonnstetter and Collura (2020), Collura and Bonnstetter (2020), Bonnstetter et al. (2018), Collura et al. (2016), Bonnstetter et al. (2015), Collura, Bonnstetter et al. (2014), Collura, Zalaquett et al. (2014), and Bonnstetter et al. (2012).

In general, the Gamma for Ipsative Validation using Electroencephalography (GIVE) process accesses asymmetric gamma wave bursts in the prefrontal cortex to validate the underlying preconscious decisions behind these self-report responses at the very moment of decision-making. The process uses asymmetric wave analysis resulting from stimulus to validate the underlying mental decisions behind these reported responses, at the very moment of decision-making, thus exposing the true thoughts behind the responses and documenting potential abnormalities between their preassessments and their actual brain activity. This process provides evidence that an evoked emotionally laden response results in corresponding brain activity and documents both the intensity of human emotional responses and the directionality of the response.

Our overall motivation has been to establish a bridge between internal subjective feelings and thoughts and the electrophysiological brain activity information that can be related on a moment-to-moment basis. We sought a method that did not require repeated trials, so that we could study complex decision-making processes in single trials. The methods described here were developed with the goal of quantifying the brain processes that underly stimulus-response activity in the brain with regard to concepts (words) and their relationship to decision-making. The intent has been to produce the ability using single trials to identify the locations and magnitudes of brain responses, and to put them in the context of an operational model in the external world. Physiological data are acquired in the form of EEG gamma activation levels (amplitudes) derived from the sLORETA localization of scalp data into Brodmann Areas 11 and 44, on left and right sides. This provides 4 values which can be quantified on a continual basis. Data are reduced to current-source-density values every 125 milliseconds and presented as both image and as quantified numeric information.

Our model is based upon identifying brain emotional responses as being of two levels, those of primary, and of secondary emotional response. An example from vision will be helpful. In vision, Brodmann’s Area 17, located at the visual pole, subserves visual sensation, the simple response to light, simple shapes, and so on. Areas 18 and 19, which receive input from Area 17 as well as other brain regions, perform visual perception in which the meaning of the sensation becomes processed, and more complex objects or events might be processed. In a similar manner, with regard to emotional processing, primary emotional sensor response is the direct sensation of a positive (pleasant) or negative (unpleasant) experience and is mediated by Brodmann areas 9, 10, and 11 at the frontal pole. Secondary emotional perception is based upon further analysis, not unlike visual perception, but in regard to the feelings and judgments about them. That is to say, emotions are processed and perceived, or comprehended, in these areas, Brodmann areas 44, 45, and 46.

The technical capabilities of the real-time sLORETA implementation provide for the ability to produce complete current-source-density images of all 6239 gray-matter voxels every 125 milliseconds, in any defined frequency bands. This method does not use z-scores or a normative database, because all data are presented and analyzed as raw current-source-density amplitudes. The frequency band chosen for these studies is the gamma band defined from 35 to 45 Hz. The 10-Hz bandwidth provides the ability of the filtering to respond accurately to brain microstate transitions on the order of the 125-millisecond frame rate. Fig. 1.1 illustrates the method used to convert an sLORETA gamma activation image to a simple code that reflects the microstate content. To estimate the activation of a given region of interest, we compute the average instantaneous amplitude of all the constituent voxels, as current-source-density in the gamma band. The computation is performed every 125 milliseconds using quadrature digital filters, which have sufficient transient response to reflect momentary changes in the voxel amplitudes that define the microstates.

Figure 1.1 Illustration coding a frontal gamma activation image into GIVE code to represent associated brain state. (Left) frontal gamma activation image for 1 subject, (center) Brodmann Areas 11 Left and Right only, and (right) Brodmann Areas 46 Left and Right only.

For illustration, the image is shown first complete on the left, then selecting only Brodmann Area 11, left and right for inspection, and then only Brodmann Areas 46 left and right. The colors of the individual Brodmann Areas determine whether the region is considered active or not, red or orange being active, and green or blue inactive. The inner regions (Areas 11) provide the two inner bits of the code, and the two outer regions (Areas 46) contribute the outer bits. The code thus represents, respectively right dorsolateral (secondary=Rs), right orbitofrontal (primary=Rp), left orbitofrontal (primary=Lp), and left dorsolateral (secondary=Ls). The code can thus be summarized as Rs Rp Lp Ls.

As the subject responds to presented information such as words or images, the system produces 8 images per second, reflecting the instantaneous changes in brain activation, within a single trial. There is not averaging or other damping applied to the data other than the digital filtering and projection into the sLORETA brain space. This provides an event-related potential measurement that is instantaneously visible from single presentations, and reflects brain activation responses to stimulus both in real-time for neurofeedback, and as image sequences for analysis.

When inspecting frontal gamma activation images in response to words that are presented by a computer screen, it is possible to identify a range of activity, such as shown in Fig. 1.1.

Fig. 1.2 presents an example of frontal lobe gamma asymmetry with positive, neutral, and negative responses. The orientation of the brain is facing forward such that the right hemisphere is on the left side of the image. The red indicates an increase in gamma activity, the blue indicates a decrease in activity, and the green indicates little or no activation. In addition to color depicting a range of response intensity, the gamma burst location is also key to interpretation. A left frontal lobe flare is an indication of acceptance or a positive response and a right-side flare demonstrates avoidance or a negative response to a stimulus.

Figure 1.2 Successive gamma activation images for one subject, for 1 s, images presented 125 ms apart.

This method has been used in a variety of settings, including responses to emotionally charged words, words in different foreign languages, faces, words relating to soft-skills, validation of forensic testimony, addiction, and smoking cessation. In these studies, it is clear that there are systematic microstate transitions that are related to the emotional and decision-making responses to each stimulus. We were motivated to find a method to systematize these transitions and put them into the context of the individual subjective experience. It was found that if we limit our analysis to a specific set of sLORETA-based regions of interest (ROI’s) that we could capture the overall appearance of the brain activation response image in a small set of values. For this work we chose Brodmann Area 11 and Brodmann Area 46, both left and right, to produce 4 current-source-density values for further analysis.

It has been found possible to classify the activation amplitudes from these four brain areas and their corresponding brain decision-making roles into a set of 16 possibilities, each representing a particular state of emotional sensation and of emotional perception. These define 16 possible microstates. We were interested to know if all possible microstates are observed in the EEG, and if they depend on task or information presented. To determine the microstates that are present, one available technique is that of clustering.

Fig. 1.3 is GIVE bubble diagram representing all possible states, in a hierarchy, with increasing content (vertical) and with negative and positive tone (horizontal). Shaded states are those appearing in the resting-state model by Custo et al. (2017) (Table 1.1).

Figure 1.3 Emotional decision model plus resting-state components.

Table 1.1

III Electroencephalography resting-state networks overlap with emotional decision-making model

Custo et al. (2017) used k-means clustering and found that it was possible to characterize the resting-state EEG with a large number of sensors with only 7 microstates, using a Meta Criterion to optimize clusters. This demonstrates that, in a first approximation, the human brain only occupies these 7 basic regions of state space, when in a resting condition. The specific states that appear are particularly interesting. When categorized using a forced-choice approach, they create a symmetric set as a subgroup of all possible states, in a revealing manner. Of these states, four in particular, their States D, E, F, and G, appear to correspond to brain activation patterns in which frontal lobes have some involvement, resembling specifically our states 0011, 0110, and 1100. Our interpretation is that these four states are included in those that a resting brain will occupy, as a matter of course without tasks or stimulation. These four resting states therefore, should be expected to be found among those in a cluster analysis of our data. In addition, we expect to see additional states associated with the stimulus presentation, as the subject’s transitions in state space become affected by specific brain activity related to emotions and decisions. The question thus arises what happens in the EEG when an individual is undergoing some task or being presented with emotionally or informationally charged information. In our investigations, we systematically presented material relevant to the subject while recording EEG and producing frontal gamma activation maps.

We have also applied k-means clustering to EEG data gathered during test conditions, using the sLORETA-derived gamma (35–45 Hz) current-source-density magnitude as a metric. We chose Brodmann’s areas 11 and 46, left and right, as indicators for relevant cortical processing. We chose n=16 for the clusters, and watched for clusters that appeared to be repeats, or overlaps, as evidence that this number was sufficient. In all cases studied thus far, less than 16 clusters were sufficient, as evidenced that generally some state would appear in 2 or more clusters. In such cases, one cluster would be observed to be of a lower amplitude, but a similar profile, to its matches. This showed that the individual might occupy the same state with larger or smaller amplitude generally, reflecting the intensity of the state, which can be of a graded nature.

We applied a size 16 k-means clustering to random data, as a way to explore the behavior of this measure. It was found that this clustering produces a clearly distributed set of all possible states, in proportions that relate to the likelihood of each state. That is, the outcomes match exactly what one would expect if the process was being applied to an honest coin toss or toss of the dice. In this case, it is revealing to examine which of the 16 possible states emerge, when real EEG is applied to data acquired during task.

IV Tobacco example

While the original use of the protocols focused on exposing decision-making pathways related to self-reporting survey responses, exposing a client to any set of words or images, especially those that may have some level of emotion, will result in frontal gamma asymmetry.

Fig. 1.4 documents the mental processing of a 48-year-old client who stopped chewing tobacco after 18 years of steady use. Data were collected in intervals of 1 month after quitting, 1 year and lastly at the three-year point. The Trigger word Tobacco was embedded in a longer series of terms so he was not predisposed to respond. Each of the stimuli was on the screen for 1.5 seconds followed by a random blank screen of 2 to 5 seconds. Fig. 1.4 shows the linear mental processing at 125th of a second over the second and a half exposure.

Figure 1.4 Frontal Gamma Asymmetry Response to Tobacco.

Fig. 1.4 is a summary of all responses from subject in tobacco cessation study, 1/8 second per image: (top row) after 1 month, (middle row) after 1 year, (bottom row) after 3 years.

Even a quick overview of Fig. 1.4 images provides a number of insights regarding mental processing changes over time, including a major change in intensity from 1 month to 1 year and then to 3 years. But even more is revealed when the images are overlaid with the bubble graph.

Figs. 1.5, 1.7 and 1.9 show the 1/8th of a second sLoreta image progressing as well as the corresponding emotional decision model cell values. Even though the images were collected over a 3-year period, the image processing, including damping settings, are identical. Figs. 1.6, 1.8 and 1.10 overlay the three image sequences onto the bubble-chart version of the emotional decision-making model.

Figure 1.5 A 1-month after stopping the use of tobacco.

Figure 1.7 A 1-year after stopping the use of tobacco.

Figure 1.9 Three years after stopping the use of tobacco.

Figure 1.6 A 1-month emotional decision model.

Figure 1.8 Subject’s state transition diagram after 1 year.

Figure 1.10 A 3-year emotional decision model.

V One-month observations

Fig. 1.5 shows the response to the word tobacco one month after cessation of tobacco use. It shows a complex sequence of states commencing with a full-brain activation in gamma, lasting for over 500 milliseconds. Afterward, there is a pulsing response with sustained activity on the right side, with no two images looking essentially alike. This reflects a complex sequence of cortical activations, including mixtures of feelings and judgments related to tobacco, mostly negative.

VI One-year observations

After 1 year, the response to tobacco is very different. Upon inspection of Fig. 1.7, it is seen to consist of two repeats of a sequence occupying about 750 milliseconds, and consisting of an initial modest bilateral activation, followed by a full activation that lasts less than 250 milliseconds. It gives the appearance of two repeated measures of the same rhythm, if looked at in musical terms (Fig. 1.8).

VII Year-three observations

At 3 years the response to tobacco is considerably reduced and simplified (Fig. 1.9). It consists of a single uniform activation, with each successive frame being the average of its neighbors, revealing a smooth, relatively slow process. It takes the full 1–1/2 seconds to complete the full sequence, which goes simply from an un-activated state, to one in which both mesial frontal poles are activated in an equal amount. Subjectively, this would reflect a much slower, simpler, and less agitating response that has been achieved after 3 years of abstinence from chewing tobacco. Moreover, the increased presence of state 0110 (Custo’s D) can be interpreted to reflect normalization of the brain, as the individual’s response to tobacco becomes both less intense in magnitude and more neutral in polarity. Visual inspection also confirms that the subject’s response is going from a state of high entropy (disorder) to one of lower entropy. This shift reflects a reduction in complexity, with the possible interpretation that the brain is doing less work now to more efficiently manage the response to a previously emotionally charged subject (Fig. 1.10).

Table 1.2 shows a summary of the transitions observed over the three recordings, categorized by the code that describes each state. There is a clear progression from left to right, comprising a cessation of the initial 1111 hyperactivated state, and a transition to being able to occupy an un-activated state, combined with states 0110, 1001, and 1110.

Table 1.2

VIII Summary of microstates observed during Emotional Decision Task and from random data

Fig. 1.11 shows the composition of the clusters derived from the k-means clustering with n=16. The EEG data are taken from the entire 1–1/2 second segments (12 samples per segment) following stimulus presentation. A total of 10,914 samples are included in this cluster analysis. It is observed that Custo et al. (2017)’s resting state D, our state 0110, occurs in four different clusters, with different overall magnitudes. The other state that occurs frequently are 0011 and 1100, which appear as Custo et al. (2017)’s states E and G. This confirms that the resting states identified by Custo et al. (2017) are also observed when using a similar method, and applied to data reflecting a wider range of mental states.

Figure 1.11 k-means clusters of states observed overall during a GIVE testing session.

Fig. 1.12 shows k-means clusters derived from uniformly random data in place of actual EEG results. The resulting clusters demonstrate that the method will nominally identify 16 unique clusters, and that these map to the 16 possible states that we identify for our modeling. That is, every combination of activation bits occurs in this sample, from 0000 through 1111, with no duplications. This confirms that the method is capable of nominal performance as expected when no patterns exist, and can be expected to produce valid clusters when patterns are imposed on the data (Table 1.3).

Figure 1.12 k-means clusters from random data 100,000 samples, normal distribution, center 2.5, SD=1.0.

Table 1.3

IX Conclusion

We have demonstrated that brain microstate analysis can be used to identify and quantify individual responses to material that has emotional or contextual meaning. We have described a model for human brain microstates that extends the concept of microstates from resting states to states associated with emotions and decision-making. It is shown that the instantaneous processing of words presented on a screen produces a stereotypical response that can be captured in sLORETA gamma activation measures taken 8 per second, for 1–1/2 seconds following the stimulus. This provides a promising avenue for the study of individual responses to emotionally charged material.

The method of measuring instantaneous magnitudes, and applying k-means clustering to the resulting distributions appears to be a useful and robust method for assessing the states that exist, and their relationships in time.

Acknowledgments

We would like acknowledge Dustin Hebets for technical contributions, and Chris Heald for contributions in the identification and implementation of k-means analysis, and interpretation of the results.

References

Baehr et al., 1997 Baehr E, Rosenfeld P, Baehr R. The clinical use of an alpha asymmetry protocol in the neurofeedback treatment of depression: two case studies. Journal of Neurotherapy. 1997;2:10–23 https://doi.org/10.1300/J184v02n03_02.

Baehr et al., 1997 Baehr E, Rosenfeld P, Baehr R, Earnest C. Comparison of two EEG asymmetry indices in depressed patients vs normal controls. International Journal of Psychophysiology. 1997;31:89–92 https://doi.org/10.1016/s0167-8760(98)00041-5.

Bonnstetter et al., 2015 Bonnstetter, B. J., Bonnstetter, R., Hebets, D., & Collura, T. F. (2015). Validation process for ipsative assessments. U.S. Patent No. 9,060,702. Washington, D.C: Target Training International, Ltd.

Bonnstetter and Collura, 2020 Bonnstetter RJ, Collura TF. Brain activation imaging in emotional decision making and mental health: a review-Part 1. Clinical EEG and Neuroscience. 2020;52:98–104 https://doi.org/10.1177/1550059420916636.

Bonnstetter et al., 2012 Bonnstetter, R., Collura, T., Hebets, D., & Bonnstetter, B. J. (2012). Uncovering the belief behind the action. NeuroConnections, Winter, 20–23.

Bonnstetter et al., 2018 Bonnstetter R, Gehrig E, Hebets D. Response process validation protocol using neurophenomenological gamma asymmetry. NeuroRegulation. 2018;5(3):93–102.

Bonnstetter et al., 2015 Bonnstetter R, Hebets D, Wigton NL. Frontal gamma asymmetry in response to soft skills stimuli: a pilot study. NeuroRegulation. 2015;2(2):70–85.

Choi and Pak, 2005 Choi BCK, Pak AWP. A catalog of biases in questionnaires. Preventing Chronic Disease. 2005;2(1):A13.

Coan and Allen, 2003 Coan JA, Allen JJB. Frontal EEG asymmetry and the behavioral activation and inhibition systems. Psychophysiology. 2003;40(1):106–114 https://doi.org/10.1111/1469-8986.00011.

Collura and Bonnstetter, 2018 Collura TF, Bonnstetter RJ. Physiology and methodology for event-related frontal gamma brain activation imaging. International Journal of Psychophysiology. 2018;131:S49 https://doi.org/10.1016/j.ijpsycho.2018.07.149.

Collura and Bonnstetter, 2020 Collura TF, Bonnstetter RJ. Brain activation imaging in emotional decision making and mental health: a review—Part 2. Clinical EEG and Neuroscience. 2020;52:105–113 https://doi.org/10.1177/1550059420916642.

Collura et al., 2014 Collura TF, Bonnstetter R, Zalaquett C. Neurocounseling: Bridging brain and behavior. Counseling Today. 2014;57(6):24–27.

Collura et al., 2016 Collura TF, Wigton NL, Zalaquett C, Chatters-Smith SS, Bonnstetter R. The value of eeg-based electromagnetic tomographic analysis in human performance and mental health. Biofeedback. 2016;44(2):58–65.

Collura et al., 2014 Collura TF, Zalaquett CP, Bonnstetter CJ, Chatters SJ. Toward an operational model of decision making, emotional regulation, and mental health impact. Advances in Mind-Body Medicine. 2014;28(4):18–33.

Custo et al., 2017 Custo A, Van De Ville D, Wells WM, Tomescu MI, Brunet D, Michel CM. Electroencephalographic resting-state networks: source localization of microstates. Brain Connectivity. 2017;7(10):671–682 https://doi.org/10.1089/brain.2016.0476.

Davidson, 1992 Davidson RJ. Anterior cerebral asymmetry and the nature of emotion. Brain and Cognition. 1992;20:125–151.

Dawson et al., 1992 Dawson G, Klinger LG, Panagiotides H, Hill D, Spieker S. Frontal lobe activity and affective behavior of infants of mothers with depressive symptoms. Child Development. 1992;63(3):725–737.

Harmon-Jones and Allen, 1997 Harmon-Jones E, Allen JJ. Behavioral activation sensitivity and resting frontal EEG asymmetry: covariation of putative indicators related to risk for mood disorders. Journal of Abnormal Psychology. 1997;106(1):159 https://doi.org/10.1037/0021-843X.106.1.159.

Ivey, Ivey, & Zalaquett, 2022 Ivey A, Ivey MB, Zalaquett CP. Intentional interviewing and counseling: Facilitating client development in a multicultural society 10th ed. Independence, KY: Cengage; 2022;41051.

Kandel, 2013 Kandel, E. R. (2013). The new science of mind. The New York Times. SR12. Accessed 30.06.14.

Kelley et al., 2017 Kelley NJ, Hortensius R, Schutter DJ, Harmon-Jones E. The relationship of approach/avoidance motivation and asymmetric frontal cortical activity: a review of studies manipulating frontal asymmetry. International Journal of Psychophysiology. 2017;119:19–30 https://doi.org/10.1016/j.ijpsycho.2017.03.001.

Matsuyoshi and Watanabe, 2021 Matsuyoshi D, Watanabe K. People have modest, not good, insight into their face recognition ability: a comparison between self-report questionnaires. Psychological Research. 2021;85:1713–1723 https://doi.org/10.1007/s00426-020-01355-8.

Piwowarski et al., 2020 Piwowarski M, Singh US, Nermend K. Application of EEG metrics in the decision-making process. Experimental and Quantitative Methods in Contemporary Economics CMEE 2018 Springer Proceedings in Business and Economics Cham: Springer; 2020; https://doi.org/10.1007/978-3-030-30251-1_14.

Sutton and Davidson, 1997 Sutton SK, Davidson RJ. Prefrontal brain asymmetry: a biological substrate of the behavioral approach and inhibition systems. Psychological Science. 1997;8(3):204–210 https://doi.org/10.1111/j.1467-9280.1997.tb00413.x.

Tomarken et al., 1990 Tomarken AJ, Davidson RJ, Henriques JB. Resting frontal brain asymmetry predicts affective responses to films. Journal of Personality and Social Psychology. 1990;59(4):791–801 https://doi.org/10.1037/0022-3514.59.4.791.

Van Honk and Schutter, 2005 Van Honk J, Schutter DJLG. Dynamic brain systems in quest for emotional homeostasis. Behavioral and Brain Sciences. 2005;28(2):220–221 https://doi.org/10.1017/S0140525X05490044.

van Honk and Schutter, 2006 van Honk J, Schutter DJLG. From affective valence to motivational direction: the frontal asymmetry of emotion revised. Psychological Science. 2006;17(11):963–965 https://doi.org/10.1111/j.1467-9280.2006.01813.x.

Wheeler et al., 1993 Wheeler RE, Davidson RJ, Tomarken AJ. Frontal brain asymmetry and emotional reactivity: a biological substrate of affective style. Psychophysiology. 1993;30(1):82–89 https://doi.org/10.1111/j.1469-8986.1993.tb03207.x.

Chapter 2

Quantitative electroencephalography (qEEG) clinical applications: a review and update

David S. Cantor¹ and Leslie S. Prichep²,    ¹Mind and Motion Developmental Centers of Georgia, Johns Creek, GA, United States,    ²BrainScope Company, Inc., Chevy Chase, MD, United States

Abstract

In the more than a decade since writing the previous version of this chapter, the development of quantitative EEG (qEEG) as a tool has greatly expanded not only for use in health care but also as an industry encompassing commercial companies offering variations of qEEG measurement sets, as well as professional Boards to provide certification and training companies to provide the professional standards to meet the requirements for certification. This chapter will serve to outline some of the multitudes of clinical applications of qEEG and its continued use in various forms of neurotherapy. As before, case studies will be used to illustrate these applications.

qEEG measurements have expanded from simple measurement derivations across the traditional broad frequency EEG bands of Delta, Theta, Alpha, and Beta to calculations within the EEG spectra in very narrow bands as short as 4 Hz and the calculation of not only a multitude of univariate measures but also complex multivariate measures that can be combined into multivariate algorithms that can describe everything from specific measures of human performance to the variations of behaviors that we have come to define a clinical disorder or syndrome.

Fundamental to qEEG are the quantitative aspects of measuring the brain’s electrical signals for neural activity and employing statistical measures of probability to derive clinical meaningfulness from these measures.

Keywords

Electroencephalography; quantitative EEG (qEEG); neuroimages; classification algorithms; neuropsychiatric disorders

I Diagnostic considerations

Quantitative electroencephalography (qEEG) providers have always asserted, appropriately so, that qEEG in and of itself is not a diagnostic instrument to qualify a disorder or syndrome no more than any one measurement tool is used in other facets of health care to make a definitive diagnostic statement of a disorder and the presence of an abnormal specific condition. Researchers in the field have attempted to provide taxonomies of brain features to try to describe the myriad of brain profiles that can occur to represent such disorders as attention deficit hyperactivity disorder (ADHD), depression, schizophrenia, anxiety disorders, dementias, autism, etc. An implicit problem with trying to match qEEG profiles to specific disorders is that psychiatric disorders are themselves heterogeneous populations and while there are some a priori qualifications (largely symptoms based) to be included in a disorder, there are variations in any sample of depressed patients or schizophrenic patients, etc. This also potentially explains the variations within a psychiatric type to their varied response to treatments. Thus, variations of brain profiles within a psychiatric subgroup might be the taxonomic method to derive an objective and meaningful portrayal of dysfunction rather than the nominal level used by the Diagnostic Statistical Manual (DSM) to define human behaviors.

Several efforts have been made to provide a way to categorize qEEG brain profile feature sets associated with medications present or for clinical conditions with specific features present (Johnstone et al., 2005). qEEG offers a powerful application tool as a method for providing convergent evidence in the identification of clinical syndromes for individuals. Over the years, various clinicians using qEEG have attempted to establish brain maps that correspond with specific disorders such as learning disabilities, ADHD, chronic alcoholism, depression, etc. While certain features may be associated with general types of disorders, the utilization of univariate features has, to date, been unable to provide unique solutions in defining specific psychiatric disorders (Coburn et al., 2006; van Tricht et al., 2014).

However, while no single feature can be diagnostic of a disorder, it has been shown that different disorders have different patterns of abnormal features. Fig. 2.1 provides examples of the presence of abnormalities (color codes in the images) the pattern of these abnormal univariate features are different for the different disorders (rows in Fig. 2.1) as discussed in Prichep and John (2017).

Figure 2.1 Group average topographic images of selected neurometric variables (columns) across different clinical populations (rows). Columns, left to right: (1) absolute power (microvolts squared) in the alpha frequency band; (2) relative power (percent) in the theta band; (3) relative power (percent) in the beta band; (4) interhemispheric coherence (percent synchronization) in the alpha band; and (5) mean frequency of total spectra of the EEG (Hz). Left panel and right panel rows show different diagnostic groups and the n in each group The color scale is proportional to the significance of the deviation from age-expected normal values, with black, no departure; blue to light green, deficits; and red to yellow, excesses. Statistical significance of entry for any group can be estimated by multiplying the value of the indicated hue on the Z-score color bar by the square root of the group n. Reproduced with permission from Prichep and John (2017).

A problem with feature matching by univariate maps of datasets is that specific metrics and enough different profile sets need to be generated to provide sufficient statistical sensitivity, specificity, and discriminability in defining these disorders. Ultimately, we also need to see how such discriminability or subtypes have meaningfulness. We will discuss treatment considerations later.

Research utilizing multivariate measure sets offers a more concise manner to capture the range of brain functional sets that best describe the probability that an individual becomes classified within a specific disorder. Fig. 2.2 from Prichep and John (1992) illustrates a way to conceptualize multivariate discriminant algorithms that can describe a disorder or for that matter even specific human functions.

Figure 2.2 Schematic of a Z-matrix of abnormal neurometric values (marked with an X) for a theoretical patient with Disorder 1. Rows of the matrix are Z-transformed features, and columns are brain regions. The Profile for the disorder is the sum of the cells (feature x region) that describes Disorder 1. Reproduced with permission from Prichep and John (1992).

The figure shows that it is possible to derive a complex algorithm that encompasses sufficient univariate features with weight to describe a significant member inclusion in a certain disorder (e.g., 95% of all Depressives would be classified using this profile disorder).

We have posted some of this earlier work in our previous edition and are doing so again here in Fig. 2.3, which provides the discriminant accuracy in identifying the statistical likelihood of a patient belonging to a particular diagnostic group (Prichep and John, 1992).

Figure 2.3 Summary of discriminant results for two groups (top panel) and multiple groups (bottom panel) Neurometric qEEG discriminant functions. The initial discriminant accuracy is indicated first (X) followed by the accuracy in the independent replication (Y), shown as (X/Y).

In Fig. 2.3, for example, in the independent replication, normal versus depression is accurate in identifying normal as normal 83% and depression as depression at 93% accuracy. Similar findings have been noted for discriminating not only normal versus a psychiatric disorder but also discriminating between or within (e.g., unipolar vs bipolar depression) clinical disorders as well.

While qEEG measures can be used to statistically match complex feature sets to disorders, other methods of qEEG analyses have become progressively more prominent in the form of variable resolution electromagnetic tomographic analysis (Bosch-Bayard et al., 2001) and a progression of methods using low-resolution electromagnetic tomographic analysis (LORETA) (Pascual-Marqui, 2002), to standardized LORETA (Pascual-Marqui, 2002), and methods of exact LORETA [eLORETA] (Faes et al., 2021). These and similar methods utilizing source localization methods (swLORETA, Palmero-Soler et al., 2007`) have emerged as commercially available tools enabling the clinician to examine brain functional measures with a reasonable spatial resolution (5 mm) and temporal resolution (up to 0.4 Hz) that is superior to current methods of functional magnetic resonance imaging (fMRI). Using special weighted functions mathematically defined, the source localization methods have also been able to target such deeper structures as the amygdala, thalamus, and nuclei in the cerebellum. These methods provide the mathematically most probable underlying sources of the scalp recorded data. Like methods using phenotypic univariate brain profiles or multivariate dominant functions, source location derived regions of interest in the brain posing as statistically deviant from age-matched normal populations metrics do not confirm or disconfirm a neurological or psychiatric disorder but can confirm brain dysfunction in terms of regions of interest and/or connectivity measures that can be matched to a patient’s presenting symptom or neuropsychological profile (Shulman & Goldstein, 2014). There is a growing number of clinicians worldwide who have been using neuroimages and derived measures to match these symptoms of performance measures to confirm the presence of underlying brain dysfunction as a consideration to explain other clinical neuropsychiatric findings.

II Forensic considerations

Arguably, the admission of qEEG in court proceedings to support evidence of impairment goes a long way to add a level of credibility and the number of cases in which qEEG has been entered into civil and criminal cases has been increasing. In civil cases, the argument is less tentative as the localization of brain dysfunction can be cross-validated by other methods of neuroimaging, both structural as well as functional. The following Case 1 illustrates how sLORETA findings can provide confirmation of an underlying neurophysiological dysfunction that corresponds not only to neuropsychological findings but other neuroimaging functional findings to validate and elucidate the presence and severity of a patient’s clinical conditions.

Case M was in a car accident on November 29, 2012. Her car was hit from behind, which caused her car to hit the vehicle in front of her. She did not lose consciousness. She described feeling like being in a fight and flight emotions for two days. She has been struggling with cognitive, and behavioral concerns related to poor attention (inattention, verbal inhibition, and vigilance) and executive function (initiation, planning/ organization issues, working memory, and self-regulation). M has hypersensitivity to noise and light. Her twin sister also stated that her sister could multitask effortlessly prior to the accident but could no longer. Neuropsychological Evaluation revealed:

• Average for age IQ

• Executive function problems

• Decreased attention

• Decreased processing speed

• Emotional dysregulation

Multiple neuroimaging techniques provided convergent findings to support damage to the brain as a result of this injury.

These findings are illustrated in Fig. 2.4:

Figure 2.4 Illustration of the concordance between 3-D imaging techniques of volumetric MRI, SPECT Scans, and sLORETA demonstrating common brain regions of interest with brain damage/dysfunction. Taken from Cantor et al. (2018).

These findings of deviation in structure and function in the same areas provide credence to brain problems that can be correlated to the symptoms reported and clinical neuropsychological findings, particularly of emotional dysregulation.

This case also illustrates the potential use of such qEEG measures in forensic diagnostic considerations as a correlative measure to support proposed brain dysfunction as more likely than not underlying functional limitations or aberrant behaviors as a course of a forensic application.

Thatcher et al. (2003) have argued that the use of norm-referenced qEEG meets the Daubert standards for admissibility of evidence in a court of law. There are four criteria needed to meet the standard.

1. Able to test hypothesis regarding expected function - qEEG measures can be fit to a normal distribution, one can test whether or not function in a given brain region or across brain regions differs in a statistically significant manner from expected values for age in the normal population.

2. Can establish a potential error rate - Since qEEG measures fit into a normal distribution, one can derive a direct estimate of standard error around any measure within the distribution, and estimate error rates depending on where within the normal distribution a given measure lies.

3. Established in peer-reviewed publications - The validity and reliability of the qEEG methodology have been published as well as illustrations of qEEG in thousands of publications concerning behavioral sciences and neurology.

4. General acceptance by clinical professionals - The utilization of EEG in published clinical research and clinical practice has expanded greatly in the past couple of decades. Admissibility of qEEG as evidence of impairment of the brain that can be correlated to compromised function and normal behavior has also increased paralleling its increased usage. A key point to consider in these cases is that qEEG is not advocated to be a singularly definitive measure of brain functional impairment resulting from an accident or inflicted injury. However, as a complementary tool that seeks to provide an age-normed reference set of metrics to specify the likelihood of impairment and the degree of impairment, it often nicely compliments neuropsychological tools and other measures in support or to contraindicate impairment.

While support to utilize qEEG in civil cases of injury seems implicit here, the use of qEEG to be used to confirm or disconfirm brain functional profiles consistent with specific criminal-type behaviors are lacking although there a limited set of studies that have attempted to do so (Calzada-Reyes et al., 2020; Evans, 2006). A primary reason for this has to do with the fact that with incarceration comes the history of such individuals confounding the differentiation of criminal behaviors from other comorbidities. For example, often in a murder case, once captured, a murder suspect who may be deemed as psychotic is medicated following an initial psychiatric examination to confirm that the patient can be certified sufficient to partake in their defense in a court of law and to do so, medication is often prescribed, Such methods, by the nature of their psychpharmacotherapeutic mechanisms, influence brain activity that can mask any set of features or sub-features that can be adequately correlated to the prominence of specific criminal behaviors. Further, such individuals are often fraught with histories that include multiple head injuries from embattlements, use of alcohol or recreational drugs, or have confounded other comorbid psychiatric conditions that may have contributed to a lifetime of experienced challenges shaping their frustrations and discourse toward criminal behaviors, for example, ADHD, (Freckelton, 2020). As such, weeding out specific components of brain measures that can be specific to specific types of criminal behaviors are near impossible in the presence of a variety of such confounds.

Nonetheless, if a patient can have his qEEG sampled in the absence of medications and if it can be obtained within a reasonable time frame from the date of a criminal event, identification of a statistical consistency to a psychiatric disorder such as schizophrenia can be argued in the presence of other clinical correlates that a patient with paranoid schizophrenia can be predisposed to violent actions if felt to be threatened. There already have been several cases in which qEEG was entered into evidence in the presence of other evidence to support the likely presence of brain injury resulting from closed head injuries resulting from accidents (William Noyd v. Cincinnati Insurance Co., et al., Gwinnett County, GA Case# 04A-10352–6), toxic insults (Missouri Circuit Courts, Cause# 22052–09567), and other cases using qEEG measures in underlying psychiatric disorders associated with criminal behavior (Ste vs Jason Bryant -Case#2000 CR 326; Camden County, GA).

III Quantitative electroencephalography methods used in therapeutic discourse

A Psychopharmacotherapeutic considerations

For the same reasons qEEG measures can be affected by medications, qEEG measures offer the possibility of establishing specific biomarkers for successful treatment. Using qEEG, measures can be used to determine drug toxicity or adverse effects on the brain which is implicit but there has been a growing body of research indicating that such measures can predict positive responses to specific classes of medications, specific medications for various disorders (Mucci et al., 2006). One perspective that has been posited is that if we know how specific agents alter or shift the distribution of the EEG spectra frequencies, then upon identifying specific abnormal features, we can adjust the distribution to normalize. For example, if the qEEG age-normed reference z-score indicates the presence of significantly increased theta power, knowing that a pharmaceutical agent typically decreased theta power would predictably shift the abnormal theta and create a balance of other EEG frequencies in a normal direction and should yield a clinical improvement. This has been posed a Lock and Key model if using qEEG to predict improved brain function to yield improved behavior or function (Saletu et al., 2006). This concept is shown schematically in Fig. 2.5.

Figure 2.5 Schematic of the Lock and Key model for drug response. This example uses z-scores for alpha (blue) and theta (green) expressed as z-scores relative to expected normal values. Baseline (left bars) shows increased alpha and decreased theta, drug state (middle) shows the effect on the qEEG of the drug (decreases alpha and increases theta), the left bars show the effect of the drug on the baseline state in a responder, now normalized.

Prichep et al. (1993) had already illustrated this concept. This study indicated that in a population of obsessive-compulsive disorder (OCD) patients there were distinct qEEG subtypes and that subtype membership was predictive of response to selective serotonin reuptake inhibitors (SSRI) medication. Fig. 2.6 illustrates the univariate feature sets that characterized the qEEG subtypes of OCD.

Figure 2.6 Group average topographic maps for z-relative power in the delta, theta, alpha and beta frequency bands for the two neurometric qEEG clusters (middle and bottom rows) of patient with DSM-IV obsessive compulsive disorder (top row). Color-coding is proportional to the mean z-score for each cluster, in steps corresponding to those shown in the z-scale. Reproduced with permission from Prichep et al. (1993).

Fig. 2.6 demonstrates two clusters within a sample of OCD patients, one with a pretreatment profile characterized primarily by elevated theta relative power and decreased alpha power, and the other cluster, just the opposite. It was hypothesized that since the primary effect of SSRI agents on the EEG is to increase that theta relative power and decrease alpha relative power. Thus, by the Lock and Key theory, SSRI agents would yield clinical improvement for OCD patients characterized by Cluster II but not Cluster I. This is indeed what was found. Eight out of the 10 patients in Cluster I did not respond well to the use of SSRIs Whereas 82.9% of the patient’s in Cluster II did so (Prichep et al., 1993). In an independent replication at Copenhagen University Hospital, 93.3% of SSRI responders were correctly classified in Cluster 2 (Hansen et al., 2003). To illustrate this predictability using a 3-D source localization presentation of a Cluster II patient (SSRI Responder) is shown in Fig. 2.7:

Figure 2.7 sLORETA image for peak within the alpha band (10.23 Hz), baseline image premedication.

The sLORETA image seen in Fig, 2.7 illustrates elevated alpha activity in a narrow band of alpha (10.23 Hz) with a maximum deviation from normal in the cingulate gyrus. Based on the group’s analyses, utilizing an SSRI agent in this OCD patient should demonstrate a reduction in this excess alpha activity. The postmedication effect is illustrated in Fig. 2.8.

Figure 2.8 Postmedication sLORETA image for patient shown in Fig 2.7. As noted, in Fig. 2.8 there is a significant decrease in the once overactive alpha activity with a corresponding improvement in clinical condition.

As noted, in Fig. 2.8 there is a significant decrease in the once overactive alpha activity with a corresponding improvement in clinical condition.

Other studies have shown that qEEG measures of the slow wave to fast wave imbalances, often found distributed in ADHD could predict improved qEEG and improved clinical outcomes when stimulant medications such as Methylphenidate or nonstimulants was applied (Arns et al., 2018; Clarke et al., 2003; Gokten et al., 2019; Leuchter et al., 2014; Ogrim & Kropotov, 2019; Suffin & Emory, 1995).

Other studies have shown how qEEG measures can predict the effectiveness of antipsychotic medications (Galderisi, 2002). And still others in the prediction of responses to antidepressants (Bares et al., 2010). While the goal of much of the work exploring the predictability of positive response of pharmacotherapeutic agents, of perhaps even greater value for qEEG is to predict nontherapeutic results or even adverse effects. For example, Iosifescu et al. (2008) investigated frontal quantitative EEG (qEEG) as a predictor of changes in suicidal ideation (SI) during SSRI treatment in major depressive disorder (MDD). Eighty-two subjects meeting DSM-IV criteria for MDD entered an 8-week, prospective, open-label treatment with flexible-dose SSRIs and completed at least 4 weeks of treatment. All subjects were assessed with the HAM-D and all subjects had four-channel EEGs (F7-FPz, F8-Fpz, A1-FPz, A2-FPz) before treatment. During the first 4 weeks of treatment 9 (11%) subjects experienced worsening SI. Left-right asymmetry of combined theta + alpha power correlated significantly with a change in SI from baseline, even when adjusting for changes in depression severity (HAM-D-17) and the SSRI utilized. These metrics were thought to serve predictive adverse outcomes from the use of SSRIs for certain MDD patients. qEEG metrics have also been able to elucidate and predict placebo effects versus drug treatment effects. For example, Leuchter et al. (2009) noted that of the medication-treated and placebo-treated subjects, 52% (13/25) and 38% (10/26) responded. Placebo responders had lower pretreatment frontocentral cordance in the theta frequency band than all other subjects (P<.006) and medication responders in particular (P<.004). Placebo responders also had faster cognitive processing time, as assessed by neuropsychological testing, and lower reporting of late insomnia (P<.03). An exploratory examination of a multiple variable models for predicting placebo response was conducted using logistic regression, in which these three pretreatment measures accurately identified 97.6% of eventual placebo responders.

As studies continue to include higher spatial and temporal resolution parameters and to apply AI applications to track differential responses to dosages, time from administration, and may include other factors such as age and interactive effects from other meds, in the future, a simple 10-minute EEG may be the best tool to use to optimize prescriptive pharmacotherapeutics to yield better outcomes while also reducing adverse outcomes in patients opting to use medical intervention