Psychiatry is currently dominated by a one-size-fits-all, trial-and-error approach that fails many patients. Precision psychiatry seeks to replace this with treatments tailored to the specific features of patients’ mental disorders. Brainify.AI, headquartered in New York, is making precision psychiatry a reality, with an innovative approach to identifying psychiatric biomarkers derived from electroencephalograms.
Psychiatric biomarkers are a crucial tool for the differential diagnosis of broad psychiatric disorders whose symptoms overlap (as well as subtypes within diagnostic categories), and are a prerequisite for precision, patient-specific treatment plans. The search for biomarkers has typically focused on genes and other molecular markers, but has been frustrated by the biological complexity and underlying heterogeneity of psychiatric conditions. Brainify.AI is pioneering a different way forward, using proprietary artificial intelligence (AI)/machine learning (ML) to translate electroencephalography (EEG) recordings into brain-based biomarkers of psychiatric disorders and treatment response, enabling better-designed clinical trials, increased drug/treatment efficacy, and more-informed and effective therapeutic choices.
Deep-diving into EEG data
EEG recordings, which reflect the simultaneous activity of billions of neurons, have a number of advantages over other potential brain-based biomarker approaches. EEG is inexpensive, non-invasive, easy to administer, and—unlike magnetic-resonance imaging (MRI) or computed tomography (CT)—captures dynamic changes in brain-activity patterns with high temporal resolution.
Brainify.AI has shown in early proof-of-principle work that its proprietary AI/ML can extract much deeper insights from EEG recordings than other approaches, for example identifying both the sex and age of individuals with high accuracy. In applying this approach to psychiatric conditions, Brainify.AI draws on rich datasets containing millions of EEG samples and billions of data points from 15,000 patients that serve as training samples for its AI/ML (Fig. 1). The insights generated by Brainify.AI’s AI/ML fuel two core products—PlaceboInsight AI and TherapyInsight AI—that together support more cost-efficient and effective psychiatric drug development, and facilitate the move towards precision psychiatry.
Fig. 1 | Patient response to treatment. Brainify.AI helps identify patients who are likely to respond to specific drugs, enhancing the efficacy of psychiatric treatments. EEG, electroencephalography.
Precision psychiatry tools
The Brainify.AI platform addresses a common problem encountered in phase 2 and 3 clinical trials of therapies for psychiatric disorders: underpowered study designs due to high placebo response rates, which can reach up to 60% in conditions such as major depressive disorder (MDD). Placebo responders dilute the signal of therapeutic efficacy in treatment arms, meaning that trials have to be larger and more expensive to generate data with sufficient statistical power to identify genuine drug efficacy.
The Brainify.AI platform identifies EEG biomarkers of placebo responders versus non-responders, and is a powerful tool to enrich patient populations and reduce placebo-response noise enhancement due to patient selection in phase 2 and 3 trials, boosting therapeutic effect signals by up to 50% and increasing the probability of trial success.
The Brainify.AI platform cuts to the heart of precision psychiatry. Today, psychiatric patients are recruited into clinical trials, and treated by their physicians, on the basis of symptoms, not underlying etiology or phenotype, which differ across patient subtypes. This heterogeneity means that current one-size-fits-all approaches often have little effect given a patient’s specific phenotype.
The Brainify.AI platform provides the solution by identifying EEG biomarkers of high- versus low- or non-responders to specific drug therapeutics. In clinical trials of investigational drugs, this provides a tool for stratifying patients to ensure the greatest chance of detecting signals of therapeutic efficacy that are otherwise swamped out by lack of effectiveness in certain subsets.
Tailoring treatment decisions
Beyond clinical trials, Brainify.AI platform-derived biomarkers can serve as companion diagnostics to enable physicians to make treatment decisions tailored to the patient’s likely response, and avoid the costly and often problematic trial-and-error approach of trying medication after medication until finding one that works—a shift that helps drug developers differentiate their products in a marketplace overflowing with cheap but often ineffective generics. Similarly, for pharma companies developing new drugs entering an already crowded marketplace with established first- and second-line treatments, companion diagnostics support the possibility for the new agents to be prescribed to patients earlier, avoiding competition with existing drugs.
While Brainify.AI is initially focused on applying its platform to MDD, generalized anxiety disorder and obsessive–compulsive disorder, both products are applicable to a wide variety of other psychiatric conditions. Brainify.AI is dedicated to realizing the full power of EEG-derived biomarkers in precision psychiatry, across diagnostic categories and for the benefit of the millions of patients who currently struggle to find effective treatments to restore their mental health.