The next phase of the technology will combine longitudinal data, proactive risk identification and smarter referrals, while keeping clinicians firmly in the loop, says NeuroFlow’s CEO.
There remains a structural divide between behavioral health and primary and specialty care. Mental health affects outcomes across conditions – but prescribing and basic screening can become a dead end when there is no coordinated path to treatment.
“One of the biggest breakdowns is that behavioral health still operates on a separate track from the rest of healthcare,” said Chris Molaro, CEO of NeuroFlow, an analytics company that helps organizations integrate mental healthcare into traditional medical settings.
“Nearly 79% of the time, a primary care physician is the one prescribing antidepressants. We need to prevent the ‘dead end’ of prescribing as the only form of treatment beyond a simple PHQ-9 or GAD-7 screening,” Molaro said.
“The same challenge exists between specialty care and behavioral health,” he continued. “Whether you’re talking about cardiology, oncology, orthopedics or chronic disease management, clinicians routinely see the impact of behavioral health on adherence, recovery and outcomes.”
For example, heart failure patients with depression are three times less likely to adhere to treatment, which leads to poor outcomes and higher mortality rates. Yet the workflows, data systems and accountability structures are rarely designed to support coordinated care across those disciplines, he added.
Fragmentation persists
There are underlying structural causes to these clinical consequences: different reimbursement, documentation and provider networks – plus the absence of a shared longitudinal view of patient risk, symptoms and engagement.
“The problem itself isn’t just technological – it’s structural,” Molaro said. “Historically, behavioral health has been reimbursed differently, documented differently, and delivered through a fragmented network of providers that often operate outside of the systems where medical care occurs. As a result, healthcare organizations have lacked a shared source of truth around patient risk, symptom progression and engagement.
“With the emergence of AI, the industry can finally play catch-up to this archaic infrastructure model,” he continued. “When behavioral health becomes part of the longitudinal patient record, rather than an isolated episode of care, organizations can move from simply identifying needs to actually managing them. That’s where we start to see meaningful improvements in access, outcomes and total cost of care.”
From reactive to proactive care
AI can synthesize existing clinical and administrative data and connect multiple weak signals that may indicate escalating behavioral health risk before a patient reaches crisis.
“AI has the potential to fundamentally shift behavioral healthcare from a reactive model to a proactive one,” Molaro stated. “Historically, we have waited for patients to raise their hand, miss appointments or present in crisis at an emergency room before deploying resources. The opportunity with AI is to identify patterns much earlier and engage with patients along a continuum of care that ebbs and flows based on needs.
“Behavioral health risk rarely announces itself through a single data point,” he observed. “It may be reflected in a combination of factors – changes in symptom scores, declining engagement, non-adherence to medication, missed appointments, chronic disease progression, social determinants of health or utilization patterns.
“AI can help connect those signals in ways that are difficult for humans to do consistently – and surface patients who may need outreach or additional support.”
Augment, rather than replace, clinical judgment
Provider organizations need to balance the promise with caution. Behavioral health depends on human relationships and context, and leaders need transparency into model data, risk assessments, recommendations and equitable performance.
“Where AI goes next is up to us,” Molaro stressed. “Healthcare organizations need to be thoughtful about where this nascent technology fits in the care process. As a rule, AI should augment clinical judgment, not replace it.
“A predictive model may identify that a patient is at elevated risk, but it cannot fully understand the context of that individual’s life circumstances, relationships or motivations,” he added. “Behavioral health is ultimately a human-centered discipline built on highly nurtured relationships. The clinician and patient need to have mutual trust in one another – it’s something AI can never replicate.”
Organizations should also be cautious about treating AI outputs as objective truth: Models are only as good as the data they are trained on, and healthcare leaders need transparency into how risk is being assessed, how recommendations are generated, and whether those models perform equitably across different populations, he continued.
“If you get the data right, the rest will follow,” he said.
The strategic and financial cases
AI presents an opportunity for health systems and insurers to collaborate in ways they never have before.
“By building shared workflows and operationalizing data, we can begin to chip away at unnecessary costs in the system and even expedite care when absolutely necessary,” Molaro explained. “Prior authorizations could even be streamlined if we lean into this new model of care.
“We ran a study with Independence Blue Cross analyzing 13,000 members on commercial health plans and found that technology intervention in a primary care setting, when linked with care coordination, resulted in a 68% increase in likelihood of receiving outpatient behavioral health services,” he noted.
“Of that population, 33% were less likely to visit the ED for a behavioral health-related issue, and total medical costs dropped by $27.63 per member per month.”
What this really means is that patients are getting better care and the system is serving them more efficiently, he said.
“When the right care pathways are established, and care teams are working off of relevant data, patients are treated on a sliding scale, based on severity,” he said. “We have the tools to make this possible in 2026 – now it’s about the structural execution at a systemwide level.”
The emerging future model
Moving to the next three to five years, there may be a shift from episodic questionnaires to dynamic monitoring that incorporates clinical data, patient-reported outcomes, engagement, utilization and AI-driven insights.
“The organizations that perform best will evolve to make behavioral health risk assessment and personalized behavioral health referrals a routine part of care, much like we screen for blood pressure, diabetes or other chronic conditions today,” Molaro predicted.
“We’re already seeing the shift from episodic screening to continuous risk monitoring,” he continued. “Rather than relying on a patient completing a questionnaire once or twice a year as the barometer, healthcare organizations will increasingly leverage clinical data, patient-reported outcomes, engagement patterns, utilization trends and AI-driven insights to develop a more dynamic understanding of behavioral health risk over time.”
A behavioral healthcare ‘matchmaker’
So, the operational use case becomes keeping current information on provider availability, wait times, specialties and proximity so care teams can connect patients with an appropriate site of care without labor-intensive manual searching.
“AI can keep a more accurate log of behavioral health provider sites of care, what wait times are like, and deliver more specificity for patients across a wide spectrum,” he said. “Think of the technology as ‘matchmaker,’ where instead of manually having case managers call across town to different providers, a dashboard delivers the most up-to-date information on availability, condition specialty, and distance from the patient. We’re moving toward a more connected ecosystem, slowly but surely.”
The expectation is organizations will know whether referrals actually resulted in follow-up care and improved outcomes. Technology is increasingly available – the harder challenge is systemwide execution and treating behavioral health as part of longitudinal population health.
“I’m a big believer in closed-loop referral management as a standard,” Molaro said. “Today, too many behavioral health referrals disappear into a black box. In the future, healthcare organizations will expect visibility into whether a patient is connected to follow up care and whether they are actually getting better. That level of accountability and transparency will be critical as organizations take on more value-based care arrangements and are forced to reduce costs of care.
“The biggest barriers are less about technology and more about execution,” he continued. “Ultimately, the healthcare system is moving toward a model where behavioral health is no longer viewed as a kick-the-can problem to be managed after symptoms worsen.”
It will increasingly be treated as a core component of an organization’s longitudinal population health strategy – the organizations that succeed will be those that can identify risk earlier, coordinate care more effectively, and deliver the right intervention at the right time in the right setting, he concluded.
