Is Artificial Intelligence the Answer to More Accessible Neurotherapy? Automating Individualized Care to Meet the Current Mental Health Crisis

ABSTRACT

Although mental health issues and other behavioral disturbances do not always rise to the level of medical diagnostic criteria, the qEEG and neuron feed back community has demonstrated that multiple symptoms and behaviors in both clinical and nonclinical populations can be improved by EEG biofeedback. Neurophysiological changes and EEG abnormalities are often nonspecific to symptoms and expressed behaviors due to known confounds such as genetics, life experience, health status, brain injury, and now pandemic-related psychosocial stressors and neurological tissue damage from SARS-CoV2 infection. The traumatic brain injury field is familiar with this problem, widely acknowledging that no two TBIs are alike. Neurofeedback offers compelling potential to improve psychosocial and cognitive-affective functioning for millions who are suffering, without the use of medications, and the need has never been greater. However, despite a half-century of development in the field, neurotherapy has not expanded beyond its status as a boutique practice, with only about 7,500 practitioners in the United States at present. A lack of consensus regarding evaluations and protocol development has created confusion and mistrust in the scientific community and the public. Potential practitioners must navigate a steep learning curve and invest significant time and money in training, equipment, and continuing education. Current models are often dependent on complex clinical decision-making to determine which metrics are included in the feedback process. These decisions are in turn dependent on clinician training, equipment capabilities, and experience. Experienced clinicians continue to debate which failure mode in the brain should determine the feedback protocols used on any given subject. Within the last decade the complexity of protocol determination has exponentially increased as new modalities introduced various forms of external stimulation to drive brain processes or interrupt habitual circuit behaviors. Conventional models for assessing the effect of neuron feed back protocols have been insufficient to evaluate the constellation of outcomes reflecting changes in both homeostatic (internal) and allostatic (responses to external stimuli) processes, as can be demonstrated in a recent publication. Many interventions have not adequately appreciated and accounted for the complexity of the systems involved in producing any one component of EEG signal or in allowing for adequate response from a wide range of brain failure modes. A new model of delivery is emerging which can provide affordable, accessible, effective neurotherapy. This presentation will describe an artificial intelligence-driven approach that can individualize therapy on a large scale. We will discuss the evolution of prior neuron feed back paradigms and review recently published data that support the efficacy and rationale for using an integrated model of allodynamic, multinetwork neuron feed back training. These data will demonstrate that it is possible, using an algorithm-driven systemic paradigm, to individualize results within a heterogenous population of neurophysiologically dysfunctional brains..