Key Takeaways
- The problem: Behavioral health has long relied on subjective assessment and manual documentation, leaving providers exposed to claim denials, audit risk, and uneven clinical quality as payers tighten scrutiny.
- The response: Companies showcased at February's BHASe Summit are introducing quantitative measurement and AI automation, from brain-network scans that gauge how sick a patient is to agents that review charts the moment notes are written.
- The evidence: Early adopters report gains, including up to 60% fewer documentation-driven denials in one substance use setting, while independent data shows AI cuts both ways: payers use it to deny care, with some tools producing denials far above typical rates.
- The path forward: Operators and consultants converge on one principle: automate the data and the back office so clinicians can spend more time on the human relationship that drives outcomes, not less.
From EEG-guided brain scans to AI that scrubs clinical notes before billing, behavioral health vendors are introducing objective measurement and automation. Where does it stop?
In cardiology, the idea of treating every patient identically would be unthinkable. A physician orders an EKG, reads quantitative output, and tailors treatment to what the data shows. Oncology works the same way. Behavioral health, by contrast, has spent decades running largely on clinical impression: a practitioner's trained judgment, a parent's report, a progress note typed at the end of a long day.
That gap was the through-line at the Behavioral Health Summit for Executives (BHASe), where a cluster of companies pitched a common thesis: the tools that made other branches of medicine measurable can finally be brought to bear on the brain and on the paperwork that surrounds its treatment. The pitches differed in their particulars, but they rhymed. Each argued that behavioral health's reliance on the subjective is no longer a necessity but a choice, and an increasingly expensive one.
EEG-Guided Neuromodulation: Putting an Objective Number on the Brain
Spencer Vigoren, director of strategic partnerships at Wave Neuroscience, whose company took home the summit's Catalyst pitch competition, made the cardiology analogy explicit. "In cardiology and oncology, the idea of treating every patient the same would be absurd," he said. "The things that we kind of take for granted in those domains of medicine is using quantitative measures and objective data to inform the treatment. And unfortunately, in the behavioral health space, some of that hasn't necessarily translated."
Wave's approach starts in the clinic with a 10-minute resting-state electroencephalogram (EEG). That reading is uploaded to Wave's platform and run through a multifactorial decision-tree algorithm against one of the largest EEG datasets in the world, then used to generate a personalized neuromodulation protocol delivered through transcranial magnetic stimulation (TMS). Both EEG and TMS are long-established, FDA-cleared technologies; the novelty Vigoren claims is in the personalization layer that sits between them.
In a follow-up interview after the summit, Vigoren put the case for personalization in sharper terms. "In a lot of medicine, infectious disease, acute trauma, population averages work fine because the biology is roughly the same patient to patient," he said. "It's when that biology varies sharply across individuals that the average starts to miss and personalization starts to matter. The sweet spot is where you can get meaningfully better outcomes at low marginal cost to personalize. We think mental health falls into that second category, and we hope to demonstrate that our approach delivers there."
The framing Vigoren returns to is the clarity that objective measurements can bring to this category. He recounted asking clinicians at the summit how they gauge the severity of a new patient's condition. The answer, repeatedly, was that they often cannot until treatment is underway. "How do you know how sick a patient is? And they're like, we don't know," he said. His pitch is that an objective brain measure could tell a clinician, before the first therapy session, whether a patient's neural networks are functioning well enough to benefit from talk therapy or medication, or whether something more foundational needs attention first. In the follow-up interview, he was careful to cast Wave as supporting the clinician rather than supplanting one. "In our model the physician still diagnoses the patient, but Wave comes alongside them and provides an analytical aid in the form of a report and a therapeutic protocol recommendation," he said. That report, he added, is meant for the patient as much as the physician. "Many patients have suffered for years without knowing what is wrong with their brain, or why earlier treatments failed. For the first time, EEG gives them an objective, visual picture of potential brain network disruption and how it ties to the way they experience the world. We use that data to build a personalized protocol that entrains normalized network behavior and, ideally, resolves the symptoms plaguing them."
Wave's near-term commercial hopes rest on a regulatory milestone. The company's MeRT system received FDA Breakthrough Device Designation for post-traumatic stress disorder in late 2024 and has since cleared the agency's review. The company says MeRT received FDA 510(k) clearance for PTSD, as an adjunctive treatment, on June 3, 2026, with a related Category III CPT code granted in May. Vigoren, who in May had told Acuity the company expected a decision within weeks, called it a milestone for the field. "It's the first ever clearance for a biomarker-guided, precision neuromodulation platform for PTSD," he said. Acuity confirmed the clearance through public statements from Wave's leadership, though the company's "first ever" characterization and the precise timing of the CPT action could not be independently verified. Vigoren framed clearance as a demand unlock across the roughly 3,200 TMS systems already installed nationwide. The personalized-TMS model is already scaling through state-funded programs for veterans and first responders, an approach examined in earlier reporting on eTMS outcomes data across seven states.
Notably, aside from three affiliated clinics in Southern California (two in San Diego and one in Newport Beach), Wave does not operate clinics. It builds software and licenses its technology to independently operated providers, a decision Vigoren tied to focus. "We're data scientists and build software; we’re not clinical operators," he said. "There are people who are much better suited to deliver care, like physicians." That division of labor, the technology company perfecting the measurement and the clinician owning the outcome, turns out to be the same boundary nearly every vendor at the summit was trying to draw.
AI Documentation Compliance: Cleaning the Record Before the Payer Sees It
If Wave is bringing objective data to the patient, Dmitry Karpov is bringing it to the chart. The CEO and co-founder of Adentris (the Y Combinator-backed company rebranded from WorkDone Health in September 2025) spent his career automating compliance and audit work, first at Ernst & Young, before turning to medical records two years ago. His company's software sits on top of any electronic medical record and reviews documentation as it is created, checking it against compliance rules, payer requirements, and internal consistency before a bill is ever generated.
The economic target is large. Karpov estimates the broader healthcare system spends roughly $20 billion a year on manual chart review, a figure dwarfed by the cost of what bad documentation produces downstream: denied claims and litigation. His framing of the problem is that catching errors early is far cheaper than appealing them later. "Unlike billing providers or billing teams, who look at documentation before they build a claim for an insurance company, we look at that earlier," he said. "By the moment billing starts, the documentation is clean and ready to be built."
He describes the product as "a QA nurse on steroids," a virtual reviewer that flags what is wrong, explains why it matters, and suggests a fix, without inventing clinical content that is not already in the record. In a residential substance use disorder facility, Karpov said, the number of documentation mistakes fell by more than half within three months, in part because clinicians learned from the flags and stopped repeating errors. He claims that translated into as much as a 60% reduction in documentation-driven denials and a 90% cut in manual chart-review workload, with the remaining 10% attributed to imperfect inputs like handwritten or scanned notes.
Those figures are the company's own and were not independently verified. But the direction is consistent with where the wider revenue cycle is heading. Industry data for 2025 put average initial hospital denial rates around 11%, up from 9% in 2022, with behavioral health among the specialties reporting rates above 15%. Missing or invalid prior authorization alone now drives a fifth to a quarter of denials. The same administrative pressure is pushing other ABA vendors toward automation: Boost, for instance, pitched a platform to automate manual scheduling, billing, and intake at the CASP 2026 conference. Against that backdrop, software that catches documentation gaps before submission is selling into real pain.
AI and Claim Denials: The Other Edge of the Blade
The same automation that helps providers is already being wielded by the entities on the other side of the claim. Karpov was candid that the technology cuts both ways, and that payers got there first. "They are ahead of providers," he said. "They already use AI to review the claims for many years. And that's actually inspired me to look at this industry."
The evidence bears him out, and it has drawn regulatory and physician alarm. The American Medical Association has reported that some AI-enabled payer tools generated care denials at rates many times higher than typical, citing figures from a 2024 Senate committee report. In a 2025 AMA survey, roughly half of physicians ranked oversight of payers' AI-driven medical-necessity decisions among their top regulatory priorities. And in January 2026, the Centers for Medicare & Medicaid Services launched its WISeR pilot, a controversial six-year model that lets private contractors use AI to review certain Medicare prior-authorization requests across six states. CMS says the contractors are paid based on the savings they generate and are penalized if they deny care improperly; critics counter that a savings-linked payment structure creates an inherent incentive to deny. Behavioral health and SUD providers are feeling the squeeze acutely, as Acuity reported in its coverage of prior authorization's 2026 pressure point.
Karpov's prediction is an automation arms race that ends, eventually, in something better. As appeals become trivially cheap to file, he expects providers to appeal nearly everything, forcing payers to respond in kind. "Every claim will be appealed because AI makes the appealing process so easy," he said. He pointed to a specific friction that AI struggles to resolve: the hours providers spend on hold chasing claim status. An AI caller can wait on hold indefinitely, but the moment a human on the payer side realizes they are talking to a bot, they may simply hang up. "It makes this a labor intensive process on both sides," he said. "I think we will get rid of that."
His longer-term vision is structural: providers and payers sharing clinical data directly, the way he believes financial audits should eventually run continuously off accounting systems rather than in an annual scramble. "Payers will be working with data, not claims," he said. "I hope to see that in the next five years."
That friction is not abstract for ABA providers. Karpov pointed to recent turmoil in Indiana, where Medicaid's overhaul of autism therapy, including steep rate cuts and tighter documentation rules detailed in Acuity's reporting on the state's reform bulletin, has left providers spending hours on the phone untangling rejected claims. The administrative drag that automation promises to relieve is, for many operators, the defining cost of doing business.
Where the Machines Stop: Keeping Clinicians in the Driver's Seat
For all the talk of automation, the vendors and operators at BHASe were strikingly aligned on its limit. Karpov is explicit that AI will never sign the note. "AI would never replace the clinician in creating a treatment plan," he said. "Clinicians will be in the driver's seat." His more provocative claim is that automation could reshape how provider organizations are built, freeing a founder to focus on clinical quality while software absorbs much of the operational and financial load. "The technology that we are working on, and others are working on, elevates the role of clinical delivery in treatment," he said. "I think it's a good thing."
Rick and Dana Loewenstein of TeamGame Advisors, who consult for ABA and behavioral health organizations on scaling responsibly, drew the line in nearly the same place from the opposite direction. They see AI's advantages across operational and clinical functions plainly. "At the same time, we're in a human services business," Rick said. "We're human beings serving other human beings. You can't hang your hat on AI." Dana sharpened the distinction: use the technology for "data, for outcomes, and how to better the delivery of service, make it more efficient," she said, but "at the end of the day, it literally is a feel, touch, personal relationship kind of industry." That measured posture echoes what other operators have learned the hard way about common mistakes in behavioral health AI adoption.
Even Vigoren, whose entire pitch is objective measurement, was careful to position Wave as additive rather than disruptive. "By no means are we going to replace any of these core interventions that are kind of tried and true," he said of talk therapy and medication. "Where I think Wave is going to be extraordinarily synergistic" is in optimizing a patient's brain function first, so that established therapies have a better foundation to work from. The analogy, again, came from cardiology: you would not send a patient with an untreated arrhythmia out to run ten miles.
The convergence is telling. A neuromodulation company, a documentation-AI company, and a pair of growth consultants, pitching to overlapping rooms of operators and investors, arrived at the same architecture for the field's future. Push objective data and automation as deep as they can go into measurement and the back office. Then stop, deliberately, at the threshold of the clinical relationship. "There is so much innovation waiting on the clinical side," Karpov said, "and I want to see the industry adopting that more and more." The question the summit left open is not whether behavioral health will become more measurable. It is whether the field can hold that line between the data and the human while every incentive pushes the machines further in.
