Every week in the ER I meet people whose symptoms are real but invisible to every tool we have.
There’s a name for why medicine keeps missing them…and it’s not what you think:
Eroom’s Law is the observation that drug discovery gets slower and more expensive every decade, despite better tools, better data, and better computing.
Some PhDs and MBAs think it’s an economic problem.
But it’s not.
It’s a measurement problem.
We’ve been building drugs on top of disease models that don’t match the people living the disease.
Nausea that derails mornings.
Pain that arrives like weather.
Flares tied to stress, meals, hormones, sleep—but never captured in a chart.
Normal labs. Normal imaging. Normal everything…except the person’s life.
When you sit at the bedside long enough, you see the same pattern:
Medicine doesn’t dismiss these patients because it doesn’t care.
It dismisses them because it can’t measure what they’re describing.
Medicine was never built to detect the biology beneath their symptoms—the biology within their biography.
I once treated a woman who checks the “temperature of her face” every morning because it predicts whether she can keep food down.
Another feels a tight band under her ribs hours before every flare.
For decades, medicine treated details like these as curiosities.
They aren’t.
They’re physiological signals:
unrecorded, uninvestigated, and unknown.
A teenager who vomited until he burst capillaries in his eyes was told it might be anxiety.
He knew the timing, the triggers, the sequence of every episode.
He was the most accurate narrator in the room.
But the system had nowhere to store his truth.
This is where Eroom’s Law actually lives.
Not in budgets or timelines, but in the widening gap between human experience and the models we use to define and cure disease.
Drug discovery keeps getting harder because we start with the wrong assumptions.
If you misdefine the phenotype, you misdefine the biology.
And if you misdefine the biology, no amount of downstream brilliance can rescue the program.
Across medicine, the same story repeats.
People living with chronic nausea, fatigue, dizziness, pain, crashes, flares, food reactions, sensory overwhelm, hormonal swings—
all reporting patterns that are consistent, predictable, and physiological.
The problem isn’t mystery.
The problem is that our system never built a way to capture any of it.
I’m an emergency physician who has practiced across multiple health systems, cultures, and continents.
And no matter where you work, one rule never changes: pattern recognition is the core of care.
Your ability to save a life depends on catching the real signal in the noise.
And one truth becomes impossible to ignore:
The body tells the truth…if you know how to listen.
The failure in modern medicine isn’t a lack of data.
It’s a lack of the right data—
the lived, structured physiology patients report every day.
The patterns that predict flares, recoveries, and crashes long before any lab value moves.
You can’t fix Eroom’s Law with bigger datasets or sleeker algorithms.
You fix it by grounding discovery in the real human phenotype.
By collapsing the search space before the first molecule is modeled.
That’s why we built a system that starts with the human pattern itself—
the day-to-day physiology people have been forced to track in their heads—
and turned it into the foundation for how we understand, model, and ultimately treat these conditions.
Because when you begin with the real phenotype—the actual biology expressed in lived experience—everything downstream changes.
Mechanisms become visible.
Targets make sense.
Discovery accelerates.
And patients stop waiting decades for answers that should have come years ago.
If Eroom’s Law is the story of medicine drifting away from the people it was meant to serve, then the next chapter is simple:
Listen harder.
Measure what matters.
Start with the humans who’ve been screaming the truth all along.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
Most doctors burn out because they tilt at the wrong windmills.
After 20 years in EM & public health, here are the fights that actually matter—
and the ones you must abandon. 🧵
Early on, I thought thoroughness = effectiveness.
Order every lab. Scan every belly. Cover every base.
That wasn’t care—it was fear.
Real effectiveness is knowing when not to test.
Protocols are scaffolding, but the zealots who worship them like scripture are dangerous.
Patients don’t read ACLS before collapsing.
Pattern recognition keeps people alive. Protocols keep you average.
I HELPED BUILD EVIDENCE-BASED MEDICINE.
NOW I’M TEARING IT DOWN.
🧵
I published a Cochrane review before I could legally rent a car.
Ioannidis was gospel.
RCTs were scripture.
We bowed to the hierarchy of evidence like it was infallible.
We thought we were saving medicine from arrogant anecdote.
But we replaced arrogance with orthodoxy.
A system that punished instinct, curiosity, and innovation.
AI just beat doctors on paper.
85% diagnostic accuracy.
The press called it revolutionary.
Cool.
I was cleaning blood and vomit off a gurney at 2:43 AM.
🧵 What Silicon Valley still doesn’t understand about medicine:
Triage, 7:08 AM.
Room 4: “Can I just sleep here tonight?”
Room 6: smells like vomit and weed.
Room 9: found down, hypotensive, no family.
AI wasn’t there.
But I was.
It nailed textbook cases.
Diagnosed zebras. Solved puzzles.
Meanwhile, I was trying to figure out if the guy in Room 3 was drunk, septic, schizophrenic—or all three.
He was peeing blood and couldn’t name a single med.
Public health broke during COVID. I know because I was inside it.
I was a state’s Chief Physician for Public Health. I signed the orders. I saw the machinations. I lived the pressure.
This thread isn’t an exposé. It’s a reckoning.
Follow if you want clarity instead of spin.
🧵
1.I believed in the mission.
When the pandemic hit, we worked 20-hour days. I helped lead testing, vaccination, emergency response. We thought we were saving lives. Maybe we did. But not always the way we told people.
2.We weren’t lying—but we weren’t honest either.
We simplified, exaggerated, softened, spun.
“We don’t have enough data” became “There’s no evidence.”
“May reduce risk” became “Safe and effective.”
It felt necessary. Until it didn’t.