Medicine Has 10,000 Diseases and ~40 Mechanisms. It Trains for the Wrong One.
Medicine is organized around diseases. Doctors learn 10,000 named conditions — their diagnostic criteria, their standard treatments, their typical presentations. When a patient arrives, the system tries to match them to a known category. If the match is found, treatment follows protocol. If it isn’t, the patient falls through.
This works well for common, well-characterized diseases. It fails systematically for everything else.
1 The asymmetry no one talks about
How many fundamental biological mechanisms underlie those 10,000 diseases?
Roughly 30 to 50, depending on how you count them:
- Inflammation and immune dysregulation
- Mitochondrial dysfunction and energy failure
- Thalamo-cortical hyperexcitability
- Autonomic dysregulation
- Vascular perfusion failure
- Biological barrier breakdown (gut, blood-brain)
- HPA axis dysregulation
- Cellular senescence
- Microbiome dysbiosis
- Epigenetic consolidation
The 10,000 diseases are not 10,000 independent phenomena. They are different combinations and expressions of this small set of mechanisms, in different tissues, at different severities, with different triggers.
A physician who deeply understands thalamo-cortical hyperexcitability can recognize it in fibromyalgia, in ME/CFS, in certain migraines, in PTSD, in treatment-resistant insomnia — without needing a specific protocol for each. A physician trained only by protocol needs a separate entry for each condition, and produces nothing when the presentation doesn’t match a known entry.
2 What this means in practice
Consider a patient who has a polysomnography. It shows alpha-wave intrusion in deep sleep (N3) — a pattern in which waking brain activity contaminates slow-wave sleep, preventing restorative function. The quantity of N3 is normal. The quality is not.
This pattern has been documented in fibromyalgia research since 1975 (Moldofsky et al. 1975). The mechanism is reasonably well understood: abnormal thalamo-cortical excitability causes the oscillation frequency to remain in the alpha range (~10 Hz) instead of transitioning to delta (~1 Hz). The result is sleep that looks adequate by duration but fails at the cellular level — glymphatic clearance doesn’t happen, metabolic waste accumulates, the brain wakes exhausted.
Gabapentin acts on this mechanism directly, via calcium channel modulation in the thalamus.
No one proposes gabapentin after the PSG. Not because the data is absent. Not because the mechanism is unknown. But because the consulting neurologist is looking for epilepsy or apnea — categories with clear protocols — and alpha intrusion without a matching diagnosis is not a category. It is a finding with no case to check.
Years later, with an ME/CFS diagnosis, the same PSG data tells a completely different story. The data didn’t change. The interpretive framework did.
3 The structural problem
Medicine’s nosological organization — disease categories first, mechanisms second — is a product of history. Clinical observation preceded mechanistic understanding by centuries. Syndromes were named before anyone knew why they happened. The infrastructure (billing codes, drug approvals, training curricula, guideline committees) was built around that structure and is now extremely resistant to change.
The result is a system with three systematic failure modes:
Failure mode 1: Unknown combinations. When a patient has a presentation that doesn’t match any single known disease but is explicable as a combination of two or three known mechanisms, the system has no entry for it. The patient is either misdiagnosed, labeled psychosomatic, or sent home with “nothing found.”
Failure mode 2: Cross-specialty blindness. Mechanisms don’t respect specialty boundaries. Thalamo-cortical hyperexcitability shows up in sleep medicine, neurology, pain medicine, and psychiatry — but these are organized as separate departments with limited cross-talk. A finding made in sleep medicine doesn’t automatically propagate to the neurologist, and vice versa. The data exists; no one synthesizes it.
Failure mode 3: Guideline lag. Guidelines codify what was known when they were written. ME/CFS guidelines in many countries still reflect the understanding of the early 2000s. A physician following current guidelines is, in some areas, practicing 20-year-old medicine — not because they are uninformed, but because the system they operate in updates slowly.
4 This is not a criticism of individual physicians
A general practitioner covers roughly 10,000 conditions. No human can maintain deep mechanistic expertise across all of them. The system asks the impossible and then blames individuals when they fail to deliver it.
The problem is architectural. The system was designed for a world where diseases were mostly single-mechanism, mostly infectious, and mostly resolvable with a defined intervention. That world no longer describes the bulk of the disease burden in wealthy countries. Multisystem chronic diseases — ME/CFS, Long COVID, fibromyalgia, dysautonomia — don’t fit the architecture. They aren’t niche edge cases; they affect tens of millions of people.
5 What mechanism-first medicine would look like
A physician trained around mechanisms instead of diseases would approach a complex patient differently:
- Map the active mechanisms — not “what disease does this patient have?” but “which biological processes are dysregulated, and how do they interact?”
- Target the accessible ones — not “what is the protocol for this diagnosis?” but “which of these mechanisms can be reached with available tools, and in what sequence?”
- Synthesize across specialties — the PSG, the autonomic testing, the immunology panel, the metabolomics — as inputs to a unified mechanistic picture, not as separate departmental outputs.
This is what some specialized ME/CFS centers now do. It is not standard. It requires time that is not reimbursed, synthesis that is not taught, and a framework that is not in the guidelines.
6 The deeper implication
ME/CFS is perhaps the clearest contemporary example of a disease that falls entirely outside the nosological system’s competence. It is multisystem, mechanism-driven, without a single biomarker, without a clean specialty home, and with a research base that outpaces clinical guidelines by a decade.
Patients with ME/CFS often end up doing the mechanistic synthesis themselves — reading the literature, connecting the PSG to the autonomic data to the immunology, arriving at medical appointments with more integrated understanding of their own pathophysiology than the physician they are consulting.
That is not how it should work. But it is what happens when the interpretive framework lags behind the data.
The data is there. The mechanisms are known. The tools often exist.
What’s missing is the architecture that connects them.
Part of an ongoing series on the biology of ME/CFS and what it reveals about medical systems.