A companion to: “If We Did Split the Spectrum: What Would It Actually Require?”

Lori Hogenkamp | Center for Adaptive Stress | evostress.blog

The autism spectrum is fragmenting under the weight of its own heterogeneity.

So is the chronic illness diagnostic system. So is psychiatry. So is women’s health. So is post-viral medicine. So is the entire conceptual scaffolding around metabolic disease. We are watching the same collapse repeat itself across every domain where bodies fail to behave the way the linear medical model expects them to.

This is not a coincidence. It is a structural signal.

And we are still misreading it.

The Autism Debate Is the Loudest Current Example

Over the past several months, the autism research community has been visibly wrestling with what to do about a diagnostic category that no longer holds. The Lancet Commission proposed “profound autism” as a carved-out subcategory. The Autism Science Foundation and Profound Autism Alliance funded a formal Delphi consensus. The New York Times asked, on its front page, whether the spectrum should be split apart. Dame Uta Frith — one of the architects of modern autism cognitive science — publicly stated that she no longer believes autism is a spectrum. The federal IACC was overhauled. The Independent Autism Coordinating Committee was formed in response.

All of this looks, from inside the autism world, like a crisis in autism research.

It is not.

It is the most visible current instance of a much larger pattern — the same pattern that is making long COVID impossible to characterize, keeping endometriosis undiagnosed for seven years on average, leaving fibromyalgia perpetually contested, splintering psychiatric categories every edition of the DSM, and producing a chronic illness landscape in which the same patient routinely accumulates four or five overlapping diagnoses without any of them quite fitting.

The pattern is not domain-specific. It is structural. And once you can see it in one place, you can see it everywhere.

The Same Collapse, Repeating

Long COVID has, after four years of intensive research and billions in funding, no single biomarker, no agreed mechanism, no reliable diagnostic test, and no consistent treatment response. Patients meeting the same criteria respond to interventions in opposite directions. The condition behaves like a regulatory phenomenon, not a disease entity.

Fibromyalgia is still actively contested as a real condition four decades after it was first described. The science has not failed to investigate it. It has failed to fit it into a categorical, single-mechanism frame.

Endometriosis takes an average of seven to ten years to diagnose, despite affecting roughly one in ten women. The lesions are visible. The pain is documented. The diagnostic lag is not a measurement problem. It is a framework problem — a condition that doesn’t behave like the model expects, in a body the model wasn’t built around.

Autoimmune disease is increasingly described not as discrete conditions but as overlapping syndromes. The same patient may carry lupus, Sjögren’s, Hashimoto’s, and POTS — four diagnoses chasing one underlying dysregulation the categorical system cannot name.

Chronic fatigue syndrome, after thirty years of investigation, remains a constellation of symptoms in search of a mechanism.

Psychiatric diagnostic categories continue to split, merge, and resplit across DSM editions — not because clinicians are imprecise, but because the underlying phenomena do not respect the boundaries the manual draws around them. Comorbidity rates approach ceiling. The same patient receives different diagnoses from different clinicians, in different decades, in different countries — not because the science is bad, but because the categories themselves are not carving the system at its joints.

Metabolic disease, once treated as a straightforward consequence of caloric excess, is now understood to involve insulin signaling, chronic inflammation, autonomic regulation, microbiome composition, sleep architecture, stress physiology, developmental programming, and environmental load — simultaneously, in ways that cannot be reduced to any single mechanism. Two patients with identical BMI, identical lipid panels, and identical glucose tolerance may have radically different underlying regulatory architectures and respond to identical interventions in opposite directions.

Women’s health across the lifespan — menstrual disorders, perimenopause, postpartum dysregulation, autoimmune predominance — has been systematically under-mapped because the conditions are regulatory rather than acute, multi-system rather than localized, context-dependent rather than constant, and have been measured against a male reference body.

Medically unexplained symptoms now constitute, by some estimates, up to a third of primary care visits in developed health systems.

This is not a list of exceptions. This is the dominant burden of modern disease.

What These Conditions Share

They are nonlinear. Multi-system. Context-dependent. Developmentally shaped. Regulatorily emergent. Stress-modulated. Architecturally variable across individuals. They do not have one cause, one mechanism, one biomarker, or one treatment. They behave like what they are: complex adaptive systems phenomena, expressed in human bodies, over developmental and environmental time.

And conventional biomedical science — built for acute, single-pathway, linear, categorical conditions — was never designed to map them.

This is the sentence the entire debate keeps refusing to land on:

The model isn’t wrong. It is being used outside its domain of competence.

Acute medicine remains one of humanity’s greatest scientific achievements. Antibiotics. Surgical anesthesia. Vaccines. Trauma care. Imaging. The germ theory of disease. The molecular characterization of cancer. These represent a kind of scientific power that no other approach has matched. The problem is not that biomedical methods failed. The problem is that we generalized those methods into domains they were never designed to map — and then mistook the framework’s incompetence in those domains for the patient’s incoherence.

The Split That Actually Matters

We do not need to split the autism spectrum. We do not need to split chronic illness into more subcategories. We do not need to refine the DSM for the eighth time. We do not need new diagnostic carve-outs for long COVID or fibromyalgia or POTS or any of the rest of it.

We need to split medicine.

Not into competing institutions or rival philosophies. Into two recognized epistemological modes within a single science of human health — each suited to a different class of phenomena, each rigorous in its own terms, each insufficient outside its domain.

Acute medicine

Built for what biomedical science was designed to do. Infections. Injuries. Single-pathway diseases. Surgical conditions. Emergencies. Identifiable agents, locatable mechanisms, replicable interventions, population-level dosing, categorical diagnosis. This is the domain where linear causation works and where standardized intervention saves lives. This mode of medicine should continue to dominate where it dominates well — and it does.

Adaptive medicine

Built for what biomedical science has systematically failed at. Chronic conditions. Regulatory disorders. Developmental variation. Multi-system syndromes. Stress-shaped illness. Conditions that emerge over developmental time, vary across context, behave nonlinearly, and refuse categorical capture. This domain requires a different epistemology: complexity science, developmental systems theory, allostatic regulation, predictive processing, longitudinal individual modeling rather than cross-sectional population averages. The unit of analysis is the system, not the symptom. The reference is the individual trajectory, not the population mean. The question is not which category the patient belongs to but what their regulatory architecture is doing, why, and under what loads.

This is not alternative medicine. It is not integrative medicine. It is not functional medicine. It is not biopsychosocial medicine. Those terms have their own histories and their own commitments, some valuable and some not. Adaptive medicine is something more specific: the application of complexity science, with its full methodological rigor, to the class of phenomena that linear medical models have demonstrably failed to capture.

It is more empirical, not less. More rigorous, not vaguer. Complexity science is not the absence of methodology. It is a methodology designed for systems where the methodology of linear causation breaks down. N-of-1 longitudinal trajectories are still data. Dynamical systems modeling is still science. Allostatic load measurement is still measurement. The instruments are different because the phenomena are different. That is not a softening of medicine. It is an expansion of what medicine can rigorously study.

What Adaptive Medicine Would Actually Do

Treat individual trajectories as primary data, not noise around a population mean. The same regulatory architecture in two patients can produce opposite outcomes under different loads, and averaging across them erases exactly the signal that matters.

Use n-of-1 longitudinal modeling rather than cross-sectional categorical thresholds. The question is not whether the patient crosses a diagnostic line on a given day, but how their system moves through time, under what conditions, with what regulatory cost.

Take regulatory architecture, not symptom checklists, as the unit of analysis. Two patients may meet criteria for autism, or for fibromyalgia, or for depression, while operating from entirely different underlying systems — one inflammatory-dominant, another autonomic-dominant, another developmentally trauma-calibrated, another metabolically constrained. The surface category conceals the architecture. The architecture is what determines what helps.

Recognize that the same intervention can produce opposite effects in different regulatory configurations — not because medicine is unpredictable, but because the same input has different consequences in different systems. SSRIs that stabilize one architecture destabilize another. Stimulants that regulate one nervous system dysregulate another. Anti-inflammatories that resolve one chronic condition prolong another. This is not a failure of treatment. It is a feature of complex adaptive systems.

Build clinical training, research design, payment structures, and institutional incentives around emergent rather than categorical thinking. The current system pays for diagnosis. Adaptive medicine would pay for regulatory understanding, longitudinal trajectory mapping, and architecture-matched intervention. That is a structural change, not a philosophical one.

This is, again, the same point: the model is not wrong. It is being used outside its domain of competence. The fix is not to abandon biomedical science. The fix is to stop demanding that it do work it was never built to do.

Why This Matters Now

Because the failure pattern is accelerating. Chronic illness rates are rising across every developed health system. Diagnostic categories continue to fragment under accumulating heterogeneity. Patients are accumulating diagnoses faster than any of those diagnoses explain. Clinicians are burning out trying to apply categorical tools to non-categorical phenomena. Research funding pours into biomarker hunts that keep coming up empty. And the public conversation, lacking a framework for what is actually happening, drifts between two equally inadequate poles: confidence that the next pharmaceutical or biomarker will resolve everything, and rejection of medical science altogether.

Neither of those is correct. The truth is structural: we are using a framework designed for one class of phenomena to map a class of phenomena it was never designed for. And the fragmentation we keep witnessing — in autism, in chronic illness, in psychiatry, in metabolic disease, in women’s health, in post-viral syndromes — is the empirical signal of that mismatch, not a sign that the science is failing on its own terms.

Heterogeneity is not the inconvenience that needs explaining away. It is the evidence.

The Closer

The world wants to split the autism spectrum. The autism spectrum is only the latest category failing under pressures it was never designed to hold.

Until medicine recognizes that it has been operating two distinct sciences under one set of instruments — one suited to acute, linear, categorical conditions, and one needed for chronic, nonlinear, regulatory ones — every diagnostic category that touches a complex adaptive system will keep collapsing under its own heterogeneity.

We will keep drawing tighter fences. We will keep losing people inside categories that cannot hold them. We will keep mistaking the failure of the framework for the failure of the patient.

The split that matters isn’t inside any single diagnosis.

It is the one we keep refusing to make at the level of medicine itself.

---

This piece is the third in a series responding to a moment many of us are watching unfold in real time. The autism research community is openly debating whether the spectrum should be split. Diagnostic categories across chronic illness, psychiatry, women’s health, and post-viral medicine are fragmenting in similar ways. And underneath all of it, a question is becoming impossible to ignore: what if the framework we are using to map these conditions is the problem, not the conditions themselves?

My first piece (“The Autism Spectrum Isn’t Collapsing. The Model Is. We Just Don’t See It Yet.”) argued that the autism spectrum is breaking down not because too many people are being included, but because a categorical, linear instrument cannot describe a multidimensional regulatory landscape. The second (“If We Did Split the Spectrum: What Would It Actually Require?”) took the splitting proposal seriously and asked what a scientifically coherent version of it would demand — and showed that meeting those demands leads not to cleaner categories but forward into architectural thinking.

This third piece widens the lens. Because the autism debate is not a local fight inside one diagnostic category. It is the most visible current instance of a pattern that is repeating across every domain of medicine where bodies fail to behave the way linear models expect them to. The question is no longer whether to split the spectrum. The question is whether to split medicine itself — not into competing tribes, but into two recognized epistemological modes within one science of human health, each suited to a different class of phenomena, each rigorous in its own terms.

This is what the Evolutionary Stress Framework has been building toward for twenty years. And it may be why this moment, however painful, is also the right moment to name what has been quietly true for a long time.

The mission of the Center for Adaptive Stress is not to reject medicine, but to help medicine evolve where its current models are reaching their limits. We believe the growing fragmentation across autism, chronic illness, psychiatry, women’s health, metabolic disease, and post-viral conditions is not evidence that these patients are unknowable or incoherent. It is evidence that modern health science is encountering classes of phenomena that require different scientific tools than the ones originally developed for acute, linear disease.

Our work focuses on building the conceptual and scientific infrastructure for that transition: integrating complexity science, stress physiology, neurodiversity, developmental systems theory, predictive processing, and systems medicine into a more coherent framework for understanding health and human variation. The goal is not to replace existing biomedical science, but to expand it — so that medicine can better model, study, and support the complex adaptive systems it increasingly encounters.

At its core, the Center for Adaptive Stress exists to help shift the conversation from asking “What category does this person belong to?” toward asking “What is this system doing, under what conditions, and at what cost?”