The pattern beneath the score.

It took decades of pushing uphill for American medicine to recognize the need to measure mental and emotional suffering. Screening is everywhere now, at the annual physical and on an intake tablet or paper form, and all of it leads to a score. A score cannot tell the physician what is driving that suffering, and it cannot tell them what to do next, in the visit where that decision gets made anyway.

The gap that has barely moved

Two people can screen positive for depression or anxiety with nearly identical scores and carry entirely different illnesses. One person may be experiencing signs of burnout. Another could be showing early stages of a bipolar-spectrum illness. A person with high anxiety scores may be contending with substance use, or with problems in attention and impulse control. Deeper still, a person could have trauma sitting beneath the presentation, or a personality structure that shapes everything above it. The score is accurate as far as it goes, and it registers that the symptoms are present. Reconstructing the pattern beneath them, or pointing to the treatment that fits it, lies outside what the score was built to do.

In the largest study of its kind, patients meeting criteria for major depression presented in more than a thousand distinct symptom combinations, most shared by only a handful of people and many by a single person.¹ A separate analysis found the criteria for the same diagnosis met in 170 different ways.² A positive screen is a reason to look closer, and the guidelines say exactly that: every positive result should lead to further assessment to confirm the diagnosis, gauge its severity, and identify what else is present.³ What that closer look requires is the thing a score was never designed to supply.

The cost of missing the pattern is well documented. Bipolar illness, to take one example, is frequently misdiagnosed first, most often as unipolar depression. In a national survey of people living with the illness, most reported an initial misdiagnosis. More than a third then waited ten years or longer, and saw an average of about four physicians, before the correct diagnosis.⁴ A meta-analysis of more than nine thousand patients put the mean interval from onset to management at 5.8 years.⁵ That is the common case, and it repeats across the other conditions a single score flattens together. Each missed pattern becomes years of treatment aimed at the wrong illness and repeated trips through the system, a cost paid first by the patient and then by everyone who pays for that care.

What a psychiatrist learns to see

As a psychiatrist, a neuroscientist, and a researcher, I have cared for patients whose histories rarely arrived as a clean diagnostic story, and the same problem kept reappearing. Symptoms would accumulate, diagnostic labels would accumulate, and the pattern underneath could take years to surface. I had seen it before medicine, as the child of a mother who carried the wrong diagnosis for years, and I knew what those years cost.

Often the information was already there, spread across a history, a set of behaviors, a family story, and a treatment record that no one had assembled in one place. What was missing was a way to recognize the pattern early enough to change what happened next. That recognition is much of what a psychiatrist develops over years of clinical work: the trained ability to notice which combinations deserve attention, and to catch the moment when something that looks like ordinary depression stops behaving like ordinary depression.

Most of these decisions start in primary care

Most people who are anxious or depressed are already in primary care before they ever see a psychiatrist. Primary care physicians write most of the prescriptions for antidepressants and anti-anxiety medications for people with employer-sponsored insurance, with the share of fills running from more than half in some states to over four in five in others,⁶ and roughly half of those treated for depression are treated in primary care.⁷ The physician in that room holds the relationship, and has to decide what to ask, what to treat, what deserves a closer look, and who needs another level of care.

Sending every ambiguous patient to psychiatry is neither realistic nor helpful in a system without the specialty capacity to absorb them. As of December 2025, 40 percent of the U.S. population, 137 million people, lived in a federally designated mental health professional shortage area.⁸ The visit itself is under the same strain. Delivering all guideline-recommended care to a typical patient panel has been estimated to take nearly 27 hours in a day,⁹ and the average primary care visit runs about 18 minutes.¹⁰ The strain sits in the system, which asks primary care physicians to make numerous decisions and convey them clearly to the patient with empathy. If there happens to be a psychiatric issue, we give them none of the psychiatric context those decisions require.

What the Care Companion does today

This gap is what I built Singula to close: to take the pattern recognition that usually lives only in a specialist’s training and put it in the room where most of these decisions are made. The product is the Singula Care Companion, which works from what the patient reports and gives the physician a psychiatric Brief before the visit.

The patient provides structured information about symptoms, history, and relevant risk factors. Singula’s clinical reasoning engine applies psychiatric logic I have encoded from my own clinical framework. The engine is deterministic, so the same input produces the same structured result every time, and every step of its reasoning is open to inspection. From that output, the system produces a physician-ready Brief: the patterns worth noticing, the complexity and risk signals, the questions worth asking, the safety considerations, and a read on whether the patient may need another level of care. A controlled generative layer renders that structured output into readable clinical language, and the clinical determination belongs to the physician.

A composite from practice

A composite from my years of practice, with details changed: a woman in her thirties arrives with scores that point to moderate depression and anxiety. The intake also records stretches of days when she slept very little and still felt full of energy, a parent treated for bipolar disorder, and an earlier antidepressant that left her agitated. Each of those details changes the questions a psychiatrist would ask next, while her scores stay exactly where they were.

The Singula intake is built to collect details like these before the visit, and the Brief puts them in front of the physician, who decides what they mean for this patient.

Singula is useful when it puts the right question in front of the physician at the moment it counts, with the pattern and the risk already weighed. I make no claim that Singula diagnosed anyone correctly, and the physician decides what every finding means. Today the Care Companion runs in independent primary care practices, where four physicians have opened a Brief for more than seventy patients.

Care now

A structured psychiatric brief for the clinician, in the moment of the visit, so the patient in the room gets a clearer path today.

The near-term product has one job: to give the patient in the room a clearer path out of a single visit, whether that means staying in primary care without an unnecessary referral or reaching specialized care before months are lost on the wrong one. That is worth building for its own sake, and it is the first of two futures Singula creates.

Innovation next

A longitudinal, real-world record of mental health care at a scale psychiatry has never had, the evidence that enables matching the right treatment to the right person and bringing new treatment options to people who have waited a long time.

Every primary care visit that uses Singula produces something the field has almost never had in structured form: a record of how a psychiatric presentation looked in a real person, at the moment a care decision was being made, captured as structured data that can be compared and followed over time.

The reason that record barely exists is built into how psychiatry works. In cardiology and oncology, illness leaves objective traces in imaging, laboratory values, and pathology. Psychiatry has no blood test or scan that confirms a diagnosis, and it still rests on how symptoms present.¹¹ For those who receive a screening questionnaire before their primary care visit, a numbered score is what surfaces. What does not yet exist is a longitudinal, structured record of how psychiatric illness presents across large numbers of real people, what clinicians did in response, and what happened next. That absence limits care, and it limits discovery.

If Singula does this once, it is a clinical decision support tool. If it does it repeatedly, over time, connecting each presentation to what the clinician did and what happened next, it builds that record: a real-world evidence base for psychiatric research and treatment innovation. The data is generated by the care itself, and that record is what makes the second future possible, which is why both futures belong to one company.

Only about three in ten patients reach remission on the antidepressant they are started on first.¹⁵ One first-line choice has to fit a group of patients who share the same symptoms and carry different illnesses underneath them, and that slows research the same way it slows care. Without a way to see which kinds of patients respond to which kinds of treatments, a developer cannot see where today’s treatments run out, or aim the next one at the patients those treatments are failing.

The same record reaches past medication. The Brief captures the range of psychotherapy approaches a presentation calls for, so a researcher could eventually study which psychotherapies fit which patients alongside which medications do. And because primary care already reaches patients a major academic center never will, a network of community practices running the Care Companion could carry trial access to people who would otherwise never be routed to one.

Building that record takes longer than the first product, because a once-a-year questionnaire cannot capture how an illness moves. The changes that matter happen between visits, in sleep, behavior, medication, and life. To follow them, Singula will grow from a pre-visit tool into a way of capturing meaningful change over time while respecting patients’ privacy and fitting into their lives.

The payoff is a change in treatment itself. A field that can see, at scale, how illness presents and responds can match the right treatment to the right person and show where today’s treatments run out so better ones get developed.¹² That is precision psychiatry, still years away,¹³ and this evidence is the path to it.

All of it depends on care that is worth it on its own terms, in a visit that serves the patient and gives the organization paying for it value it can measure. The evidence follows from that care.

Why primary care, and who ultimately pays

Primary care is where this must start. That is where the patients are, and it is also the one setting where three things already sit together: a patient whose pattern is hard to read in a short visit, the visit where the treatment decision often gets made, and the system that absorbs the cost when that decision goes wrong. That alignment is already how primary care works.

That alignment lets one tool serve everyone with a stake in the visit. The patient gets a visit that starts from a fuller picture of what they are living with. The physician gets psychiatric context, and the practice gets the structured assessment and the documentation that existing behavioral health integration codes such as CPT 99484 require. Those codes also require a recurring monthly care-management service, which the practice furnishes and bills, and Singula does not provide billing or coding advice.¹⁴ The system carrying the risk gets the earliest chance to recognize a patient headed for the wrong course.

Two different buyers pay for this, at two different times. First come the practices and risk-bearing systems that put it in the visit, who carry both the visit and what follows it, with reimbursement easing adoption. Later come treatment developers and innovation partners, where a structured record of how psychiatric illness presents and responds holds value well beyond any single visit.

Learning to see

I have spent my career on one question: what does this patient have, beneath the surface? Answering it early is what lets the right treatment follow, before a patient loses years to the wrong one.

Anxiety and depression are where the question begins. They are the visible, measurable tip of the iceberg. Beneath them sit the drivers, whether trauma, temperament, family history, or substance use. Sometimes they are longstanding traits that converge with compounding stress, and sometimes they are the early signs of something more serious driven by internal biology. How those forces operate in a particular person is what determines the right care.

Seeing it at scale runs through infrastructure: a system that integrates behavioral health and supports the providers who shoulder the heaviest burden of care. Singula is built for the primary care frontline, where every well-handled visit opens both of its futures: a clearer path for the patient, and a record of how the illness looked when a real decision was made. Connected over time, those observations become an account of how mental illness moves through people’s lives and responds to what we do.

The stakes are real and the problems are solvable. A patient at risk of losing years to the wrong treatment can be recognized at the first visit. The clinician who was handed a number has something real to reason from. Further out are the people waiting on treatments that do not yet exist, whose illnesses the field could never see clearly enough to match them to the right care.