HN Debrief

I was recently diagnosed with anti-NMDA receptor encephalitis

  • Public Health
  • AI
  • Open Source
  • Programming
  • Healthcare

The post is a firsthand writeup from BurntSushi about anti-NMDA receptor encephalitis, an autoimmune disease where antibodies attack NMDA receptors in the brain and can present as panic, delusions, movement problems, seizures, and other neurological symptoms. He describes a month-long slide from flu-like symptoms and intense anxiety into psychosis, falls, and hospitalization. A primary care workup initially looked reasonable because the early symptoms were vague. The real failure came later. An ER cleared him medically and sent him to psychiatric care without a neurologist seeing him. A lucky connection got him transferred to Brigham and Women’s, where MRI findings and a positive cerebrospinal fluid antibody test led to the correct diagnosis and treatment with steroids, IVIG, and follow-up neuroimmunology care. He is recovering and joined a clinical trial.

If symptoms escalate fast, do not let a fresh psychiatric label end the workup. Keep a detailed timeline, bring an advocate, and push for neurology or other specialty review when the pattern does not fit a normal psych history.

Discussion mood

Sympathetic and a little shaken. Many people were relieved the author survived, angry at how easily serious illness gets trapped in psychiatric pathways, and newly wary of how much diagnosis still depends on persistence, luck, and having someone in the room to push back.

Key insights

  1. 01

    Psych labels create powerful anchoring

    Once generalized anxiety disorder or another psychiatric diagnosis hits the chart, later clinicians are primed to read worsening symptoms as more of the same. That matters here because the ER could lean on a recent psych framing plus a negative outpatient workup and never seriously reopen the neurological differential until a transfer broke the loop.

    Treat any new psych diagnosis during a fast physical decline as provisional. Ask what evidence would overturn it and who is responsible for checking that before disposition.

      Attribution:
    • epcoa #1
    • burntsushi #1
    • cgh #1
  2. 02

    Timelines and advocates change outcomes

    Detailed symptom chronology and a persistent spouse did more than provide emotional support. They supplied the pattern recognition that made a neurosurgeon and later a neurologist take the case seriously. Another autoimmune encephalitis patient said rapid treatment only happened because a specialist saw the progression immediately and acted within a week.

    When symptoms are evolving, write down dates, new deficits, falls, odd behaviors, and medication changes in one place. Bring that document and another person to every high-stakes appointment or hospital visit.

      Attribution:
    • burntsushi #1 #2
    • cstaszak #1 #2
    • VoidWhisperer #1
  3. 03

    This diagnosis is younger than people think

    Anti-NMDA receptor encephalitis was only described in 2007, which reframes the story from simple malpractice to a moving frontier in neuroimmunology. Commenters tied that to a broader point that biomedical progress is real but uneven. New tools and therapies are arriving fast, yet awareness in frontline care lags badly behind discovery.

    For unusual presentations, check whether the relevant disease category is recent or rapidly evolving. Frontline clinicians may be practicing behind the literature even when they are competent.

      Attribution:
    • bonsai_spool #1 #2
    • jr3592 #1
    • pibaker #1
  4. 04

    AI helps most as diagnostic scaffolding

    The strongest AI anecdotes were not about miraculous one-shot diagnosis of encephalitis. They were about structuring symptom logs, surfacing plausible syndromes like MCAS, POTS, dysautonomia, or Ehlers-Danlos, and helping patients ask better follow-up questions. BurntSushi’s own use of Gemini during the illness was mostly unhelpful and even fed self-doubt by overfocusing on medication side effects.

    Use AI to organize history, generate questions, and sanity-check whether the workup is too narrow. Do not use it to downgrade severe symptoms or accept a neat answer that fits the current chart.

      Attribution:
    • forrestpitz #1
    • keithnz #1
    • burntsushi #1
  5. 05

    Rare-disease diagnosis is a triage problem

    A neurologist pointed out that the hard part is not ignorance of databases. It is deciding when a vague presentation has earned expensive tests, specialist review, MRI, spinal tap, or immune workup. Anti-NMDA can look psychiatric early, multiple sclerosis can look similar later, and even objective tests arrive in stages rather than all at once.

    If you are building care workflows or decision support, aim at escalation thresholds rather than magical final diagnosis. The key question is when to widen the workup, not whether a static symptom list maps to one disease.

      Attribution:
    • tsoukase #1 #2
    • burntsushi #1
  6. 06

    Autoimmune encephalitis has very different trajectories

    People with lived experience of autoimmune encephalitis added that this umbrella includes multiple subtypes and wildly different recoveries. One person with LGI1 autoimmune encephalitis got treated within a week and is mostly back to normal, though still managing fatigue and cancer surveillance. Another family described NORSE with months in an induced coma and permanent disability. That range makes early recognition feel far more consequential than the headline rarity suggests.

    Do not think of autoimmune encephalitis as one condition with one prognosis. Fast specialty referral and treatment timing can determine whether this becomes a reversible crisis or a life-altering injury.

      Attribution:
    • cstaszak #1 #2
    • burntsushi #1
    • greenimpala #1

Against the grain

  1. 01

    Doctors miss zebras because they prevent stampedes

    The pushback to the anti-doctor mood was that frontline medicine sees far more harmless, self-limited, or anxiety-driven complaints than rare brain inflammation. Triage has to optimize for the base rate, and no test or imaging threshold can eliminate false negatives without exploding false positives, wait times, and incidental findings.

    Demanding maximal testing for every scary symptom is not a workable standard. The better target is faster reconsideration when the course stops matching the common-case assumption.

      Attribution:
    • WarmWash #1
    • TaupeRanger #1
    • haldujai #1
  2. 02

    Internet self-diagnosis overloads rare-disease clinics

    A specialist-facing view on MCAS argued that social media and chatbots are now sending large numbers of convinced but incorrect patients into already scarce specialty slots. The subtle point is not that these patients are faking. It is that a flood of low-probability referrals creates fatigue and can make real cases harder to reach in time.

    If you use AI or online communities to explore symptoms, do the basic workup and document objective findings before escalating to niche specialists. Otherwise you may worsen the access problem you are trying to solve.

      Attribution:
    • Aurornis #1 #2
    • randerson #1
  3. 03

    Bias is broader than misogyny framing

    Several commenters pushed back on reducing missed diagnosis to sexism alone. They argued the deeper failure is a system that overweights common explanations, underweights uncertainty, and often dismisses anyone with pain, fatigue, disability, or a prior psych label. Gender disparities are real, but the operational fix is better diagnostic process, not a narrower moral label.

    When you evaluate care quality, look for structural failure modes like anchoring, time pressure, and poor handoffs in addition to demographic bias. Those process failures often hurt multiple groups at once.

      Attribution:
    • Auracle #1
    • anotherevan #1
    • haldujai #1

In plain english

anti-NMDA receptor encephalitis ↩
A rare autoimmune disease where antibodies attack NMDA receptors in the brain, causing inflammation and symptoms that can include psychosis, seizures, movement problems, and memory issues.
dysautonomia ↩
A group of disorders where the autonomic nervous system does not regulate functions like heart rate, blood pressure, and digestion properly.
Ehlers-Danlos syndrome ↩
A group of inherited connective tissue disorders that can cause hypermobile joints, fragile tissues, pain, and other systemic problems.
ER ↩
Emergency room, the hospital department that treats urgent and emergency medical problems.
IVIG ↩
Intravenous immunoglobulin, a treatment made from donated antibodies that can help calm harmful immune activity.
LGI1 autoimmune encephalitis ↩
A subtype of autoimmune encephalitis linked to antibodies against the LGI1 protein, often causing seizures, confusion, and memory problems.
MCAS ↩
Maneuvering Characteristics Augmentation System, a flight control system associated with the Boeing 737 MAX aircraft.
MRI ↩
Magnetic Resonance Imaging, a scan that uses strong magnets and radio waves to create detailed images of structures inside the body.
NORSE ↩
New onset refractory status epilepticus, a severe condition where a person without previous epilepsy develops continuous seizures that are hard to stop.
POTS ↩
Plain Old Telephone Service, the traditional analog phone network used before internet-based telephony became common.

Reference links

Background on anti-NMDA and autoimmune encephalitis

Diagnostic references and clinical tools

Healthcare bias and system issues

Mortality and population statistics

Books and media mentioned

  • The Master Switch
    Recommended in a side discussion about how technology ecosystems become centralized and degraded over time

Other medical and science references