steve pribut

Sports Medicine, Science, Running & Fitness. Lecturer, Dept of Biomedical Engineering GWU. Asst Clinical Professor of Surgery GWU.

[steve pribut @iSteeve]

Horse head Nebula from the backyard.

Simple image processing

Seestar S50

[steve pribut @iSteeve]

@mikenevermiss Prompts are overrated.

[steve pribut @iSteeve]

An interesting approach. The first step must be to determine the number of tokens that will be eaten up by this process.

[MIKE @mikenevermiss]

Anthropic engineer:

"you're not supposed to prompt Claude. you're supposed to build a system that prompts itself [loops]."

this is one of the best workflows I've seen in a long time

in this video he breaks down exactly how most people are building loops wrong:

- the memory file you never set up, so every loop starts from zero

- the sub-agents that 95% of builders have never split apart

- the stop condition setup that keeps loops from running forever and billing you in your sleep

- why writing one prompt a day is the slowest way to use Claude

if you've been using Claude for more than a month and still typing every task by hand, you've been running one prompt when you could be running a system of loops

instead of another prompt tonight, watch this

make sure to bookmark it before it gets buried

full guide in the article below

[Eric Topol @EricTopol]

Until now, we haven't seen medical AI models intended to provide end-to-end care after presentation to an emergency department or longitudinally via 3 sequential outpatient visits. There's a lot to unpack in 2 new @nature publications today. Summarized in a new post.

[steve pribut @iSteeve]

The ChatGPT comments appearing in this thread were generated without adding the context of our previous conversations. It just recalled the details.

[steve pribut @iSteeve]

If you don’t know a topic at all, this could be good. If you know it well or even partly there are better methods and perhaps better LLMs. If you are looking for graduate level details these two prompts appear weak. I’ll try and share ChatGPT Opus’s opinion on a new search and our past searches on Mitochondrial disorders below. This is a topic I’ve interested and touched on in lectures I’ve given to graduate Biomedical engineering students the past several years, without using LLM assistance.

[Nav Toor @heynavtoor]

https://t.co/mFvSsFIzlm

[steve pribut @iSteeve]

The Stanford prompt is clever, but it is optimized for perspective-taking rather than understanding a biological system.

When I think about how you’ve approached topics over the years—from gait biomechanics to Parkinson’s disease to HERV-K to protein misfolding—the question you’re usually asking is not:

“What do five different people think?”

but rather:

“How does this system actually work, where does it fail, how do we detect the failure, and what can we do about it?”

That is a very different intellectual task.

For mitochondrial disorders, I’d be tempted to structure the entire field almost like an engineering failure analysis.

Mitochondria: What are they supposed to do?

ATP production
Metabolic integration
Calcium buffering
Apoptosis signaling
ROS signaling
Innate immune signaling
Adaptation to stress

What can go wrong?

mtDNA mutations
Nuclear DNA mutations
Defective replication (POLG)
Defective respiratory chain complexes
CoQ deficiency
Defective fusion/fission
Defective mitophagy
Defective transport of metabolites

Why do some tissues suffer more?

Brain
Muscle
Heart
Retina
Cochlea
Endocrine organs

High-energy tissues live closest to their energetic limits.

Why is diagnosis so difficult?

Heteroplasmy
Threshold effects
Tissue-specific mutation loads
Age-dependent progression
Variable penetrance

Two patients with the same mutation may look remarkably different.

What diseases emerge?

Not as a memorized list but as recurring patterns:

Stroke-like syndromes (MELAS)
Myoclonic epilepsy (MERRF)
Optic neuropathy (LHON)
Developmental neurodegeneration (Leigh)
Progressive external ophthalmoplegia
Cardiomyopathy
Diabetes + deafness syndromes

What treatments actually have evidence?

A surprisingly short list.

Strong evidence is much scarcer than many people realize.

What are the biggest unresolved questions?

This is where things get exciting.

Questions such as:

Why do some mtDNA mutations devastate one tissue but spare another?
How important are mitochondrial defects in Parkinson’s disease?
Alzheimer’s disease?
ALS?
Aging itself?
Can mitochondrial gene editing become practical?
Can we selectively alter heteroplasmy?
Can mitochondrial transplantation become clinically useful?

⸻

One reason I think this framework fits you particularly well is that it resembles the way you teach graduate biomedical engineering students.

You often move from:

Normal system → Failure modes → Measurement → Intervention → Future directions

That’s exactly how I would teach mitochondrial medicine to a room of engineers.

In fact, if you decide to do a serious review of mitochondrial disorders, I would organize it as:

Mitochondrial Biology
Genetics and Inheritance
Bioenergetics and Metabolism
Classification of Mitochondrial Disorders
Major Clinical Syndromes
Diagnostic Strategy
Acute Metabolic Crises and Metabolic Stroke
Organ-Specific Manifestations
Current Treatments
Exercise and Rehabilitation
Emerging Therapies
Mitochondria in Aging and Neurodegeneration
Open Questions and Research Frontiers

That structure would take you from Krebs cycle and electron transport chain all the way to MELAS, arginine therapy, mitophagy, Parkinson’s disease, and the future of mitochondrial gene therapy without getting lost in a catalog of rare syndromes.

And, knowing you, you’d probably spend twice as much time in sections 12 and 13 as in the rest of the book. The mitochondria become especially interesting once they stop being merely power plants and start acting as signaling, sensing, and decision-making organelles. That’s where Nick Lane, Zeviani, Kelly, neurodegeneration, and modern systems biology all start to intersect.

[steve pribut @iSteeve]

From ChatGPT, analyzing our usual approach with my questions and comments vs the Stanford Prompts:

For someone like you, Steve, I think it is often weaker than your usual approach, though it can occasionally reveal a blind spot.

The strength of the “5 experts” prompt is that it forces consideration of:

Clinical reality
Evidence
Criticism
Incentives
Historical context

The weakness is that it artificially creates disagreement even when little exists, and it can give equal weight to unequal viewpoints.

For example, in mitochondrial disease:

The Practitioner and Academic perspectives are grounded in large bodies of evidence and experience.
The Skeptic perspective is useful, but there is not a large serious scientific movement arguing that primary mitochondrial diseases are mostly misdiagnosed or nonexistent.
The Economist perspective can identify incentives, but mitochondria are not like opioids, statins, or antidepressants where huge commercial forces have shaped decades of practice.
The Historian perspective can be insightful but often becomes analogy rather than evidence.

When you ask me something like:

“Teach me mitochondrial disorders at the level of a physician-scientist. Include genetics, biochemistry, diagnosis, controversies, treatments, limitations of evidence, and future directions.”

that usually gets you closer to what you actually want.

In fact, after talking with you about:

MELAS
Leigh syndrome
mtDNA heteroplasmy
mitochondrial signaling
Nick Lane
Zeviani & Viscomi
Jeff Kelly
neurodegeneration

I think your preferred framework is different.

You tend to think in layers:

Layer 1: Mechanism

What is actually happening?

Oxidative phosphorylation
ETC complexes I-V
ATP generation
ROS signaling
Mitophagy
Mitochondrial dynamics

Layer 2: Failure Modes

What breaks?

mtDNA mutations
Nuclear gene mutations
Defective protein import
Defective fusion/fission
Defective mitophagy
CoQ deficiency
Krebs cycle defects

Layer 3: Clinical Expression

What diseases result?

MELAS
MERRF
Leigh
LHON
Kearns-Sayre
POLG disorders
Adult mitochondrial myopathies

Layer 4: Detection

How do we know?

Genetics
Muscle biopsy
Lactate
Imaging
Histology
Functional testing

Layer 5: Intervention

What can we do?

Arginine
CoQ10
Exercise
Ketogenic approaches
Gene therapy
Mitochondrial replacement
Experimental therapies

Layer 6: Open Questions

What remains uncertain?

Secondary mitochondrial dysfunction
Long COVID
Neurodegeneration
Aging
Biomarkers
Heteroplasmy thresholds
Gene editing

That structure is much closer to how a biomedical engineer, clinician, and scientist think about a field.

The Stanford/Claude prompt is excellent when exploring a topic where incentives, narratives, and competing interpretations dominate the discussion—politics, education, economics, AI safety, social science, even parts of nutrition.

For mitochondrial disorders, I’d rather ask:

“Build me a physician-scientist map of mitochondrial disease. Start from mitochondrial biology, then explain genetic mechanisms, major disease categories, diagnostic approaches, treatment evidence, controversies, and future therapeutic directions. At each step distinguish established knowledge from emerging hypotheses.”

That prompt produces something much closer to the way you have approached Zeviani, Nick Lane, Parkinson’s disease, HERVs, CAR-T therapy, and your GWU lecture material.

In other words, the Stanford prompt is a good debate prompt.

Your natural style is usually a systems-biology prompt.

For mitochondrial disease, I would choose the systems-biology approach every time.

[steve pribut @iSteeve]

The lowdown

[Rahul @sairahul1]

https://t.co/TIlBSKlpjN

[steve pribut @iSteeve]

Efficient market or effective hype?

[Rahul @sairahul1]

Apple went public at under $2 billion.
15x revenue.
1980.

SpaceX wants you to buy at $2 trillion.
100x revenue.
2026.

That is not getting in early.

That is being the exit for venture capitalists who bought in years ago at a fraction of what you are being asked to pay.

Almost none of the retail investors buying this IPO will read the 300 pages before the book closes on June 11.

I fed all 300 into AI and read them for you.

Here is what you will miss ↓

[steve pribut @iSteeve]

@albertoortegana @drkeithsiau The fastest finger prize is for the first responder. The answer could be wrong or correct.

[steve pribut @iSteeve]

Fungi or not fungi, That is the question.

[Keith Siau @drkeithsiau]

Like an enchanted forest. Where was this photo taken?

[steve pribut @iSteeve]

@drkeithsiau Looking like mushrooms, this appears to be in the colon. multiple large pedunculated polyps.

[Keith Siau @drkeithsiau]

What causes this appearance in the foot?

[steve pribut @iSteeve]

@jstein_star Do you trust the government as an intermediary to give portions of the shares to the public? And not go to our upcoming Arc de Triumph?

[steve pribut @iSteeve]

HAL Astronomy on YouTube - the most recent monthly meeting covers all you'd possibly want to know about Moon Observation and current exploration.

https://t.co/BaqBlFtowC

[steve pribut @iSteeve]

@IhabFathiSulima Marfans with pectus excavatum

[Andrej Karpathy @karpathy]

- Drafted a blog post
- Used an LLM to meticulously improve the argument over 4 hours.
- Wow, feeling great, it’s so convincing!
- Fun idea let’s ask it to argue the opposite.
- LLM demolishes the entire argument and convinces me that the opposite is in fact true.
- lol

The LLMs may elicit an opinion when asked but are extremely competent in arguing almost any direction. This is actually super useful as a tool for forming your own opinions, just make sure to ask different directions and be careful with the sycophancy.

[steve pribut @iSteeve]

Tell me that you have never read the constitution (and Bill of Rights) and have no idea what is in it without stating that.

WWTJS what would Thomas Jefferson say?

[Andrew Weinstein @Weinsteinlaw]

When a federal agency tasked with protecting all Americans decides to mandate a specific religious worldview, it ceases to protect democracy and begins to threaten it.

[Eric Topol @EricTopol]

This is the kind of work, published today, that should be fully supported and prioritized at the highest level. Currently, there is no vaccine vs the Ebola Bundibugyo virus, the source of a major outbreak https://t.co/qM6bWLtmcb

[steve pribut @iSteeve]

Ignorance and demonization of science should not prevent the development of critical vaccines.

[Eric Topol @EricTopol]

This is the kind of work, published today, that should be fully supported and prioritized at the highest level. Currently, there is no vaccine vs the Ebola Bundibugyo virus, the source of a major outbreak https://t.co/qM6bWLtmcb