umar (@umarelbably) - Twitter Profili | Zamantika Mersobahis Locabet

umar@umarelbably·41m

ZXX

0

5

umar@umarelbably·1h

i built peer (frompeer.com), largely inspired by your entire vision. you've spent years proving health decisions should be governed by evidence. i believe that evidence should be accessible to anyone for free, not just researchers and doctors. peer is a health research tool built for honest, transparent, rigorous answers: - gold standard sources only (pubmed, clinicaltrials . gov, fda, + more) - source quality weighting (rct vs mouse vs case report) - claim verification, cannot hallucinate citations - transparent methodology built for your community. would love to work with you.

English

0

2

79

Bryan Johnson@bryan_johnson·1h

Is your product legit? Consumers don’t know who to trust. We just launched a new marketplace of certified brands, starting with two of my favorites: Eight Sleep: I achieved 8 months of perfect sleep on this bad-boy. Plunge Sauna: don’t forget to ice your boys. My protocol is 200°F (93°C) for 20 min, daily. These are products I trust: + backed by research + vetted by my clinical team + studied for longevity We reject most products. If you want to be considered, get in touch: partnerships at bryanjohnson dot com

English

52

11

305

38.1K

umar@umarelbably·1h

Correct. ~90% failure rate from rodent models to human efficacy [pharma]. Has held for decades.

Parmita Mishra@parmita

Don’t laugh! Rapamycin worked in mice and not in human beings. Penicillin would fail testing in a standard tox packet but it saved human lives. The FDA said it wants mouse testing to be the exception, not the rule, in “3 to 5 years.” I wrote thousands of words about how organoids plus AI can reduce the number of mice we torture. I met Dr. Manolis Kellis head of MIT comp bio who said “instead of 1000 mice, your approach, alongside AI, could reduce it to one or two in the beginning.” I’d go as far as to say: mice and beagles won’t need to be tortured at all. My baby beagle Plato, my childhood beagle Bruno, beagles like them won’t be tortured anymore. We will make drug discovery human and it’ll be beautiful.

English

0

1

45

umar@umarelbably·2h

for anyone interested i benchmark peer two ways. 1) healthbench professional. openai's published test for medical research AI. 147 expert-graded questions, independent grader, public rubric. 2) my own benchmark. there aren't real standards yet for evaluating health research AI, so i built one. rubric and methodology adapted from the gold standards; healthbench, med-palm 2, medhallu, and the decompose-then-verify approach used in medscore and factscore. it covers supplement research, drug safety, clinical trials, multi-turn conversations, and adversarial safety-trick questions. every answer is scored on source grounding, citation quality, completeness, honest uncertainty, safety, and clarity. a factual accuracy failure gets a zero. i also break answers into individual claims and verify each one against the studies the response cited, with a fresh pubmed search for anything unresolved. i run the same questions through frontier models from openai, anthropic with web search enabled and score them the same way. peer has outperformed competitors in my internal testing. not claiming perfection, but claiming rigor.

English

0

57

umar@umarelbably·18h

OpenEvidence is a $12B company used by 65%+ of US doctors. I built the version anyone can use for free. Ask Peer a health or medical research question: frompeer.com I thought it was weird that OpenEvidence's model is pharma ads + partnerships with prestige journals; feels dangerous to miss the broader gold standard research. Peer strives for the most honest, transparent, and rigorous answers in health research: - Gold standard sources (PubMed, ClinicalTrials . gov, FDA, NHS, + more) - Source quality weighting - Claim verification (cannot hallucinate citations) - Transparent methodology

English

13

15

131

71.3K

umar@umarelbably·2h

@statnews @EvidenceOpen the now what is literally peer. access the same evidence for free, for anyone. x.com/umarelbably/st…

umar@umarelbably

OpenEvidence is a $12B company used by 65%+ of US doctors. I built the version anyone can use for free. Ask Peer a health or medical research question: frompeer.com I thought it was weird that OpenEvidence's model is pharma ads + partnerships with prestige journals; feels dangerous to miss the broader gold standard research. Peer strives for the most honest, transparent, and rigorous answers in health research: - Gold standard sources (PubMed, ClinicalTrials . gov, FDA, NHS, + more) - Source quality weighting - Claim verification (cannot hallucinate citations) - Transparent methodology

English

0

50

STAT@statnews·2h

OpenEvidence Won the Hearts of Doctors. Now What? @EvidenceOpen barged into medicine and quickly became indispensable. In just a few years, it’s drawn hundreds of thousands of clinicians to its AI-powered search, upending decades-old reference tools. Now that it’s won over doctors, where does it go next? Join Us: statnews.com/stat-summit-we… #STATBreakthrough

English

1

0

2

2.2K

umar@umarelbably·2h

correct and incorrect. all the best research ends up on pubmed eventually. peer still has live web search from a curated list of sources to catch anything it doesn't directly index. also, doctors have literally told me they shouldn't be making recommendations based on very early studies that lack PR and RCTs. the key is balance, which i believe we strike. we already are filling gaps even openevidence has failed to do: x.com/umarelbably/st…

English

0

1

6

Pratik Dalwadi@moderndas·2h

@willchen500 @umarelbably Also really good quality clinical content is in medical journals and meta review sources like cochrane review, clinical text books so at some point one needs this. Publicly available sources like pubmed are not enough because you’d miss out on other recent info.

English

1

0

44

umar@umarelbably·2h

@moderndas @willchen500 peer already pulls cochrane, uspstf, ahrq, dailymed, openfda, orange book, pubmed. all cited inline. i'd recommend trying it out for pharma related questions.

English

1

0

34

Pratik Dalwadi@moderndas·2h

@willchen500 @umarelbably Exactly my question. Openevidence announced recent partnership with cochrane review. I built a similar thing for pharmacists and it wasn’t as hard as i thought. But i need content access to clinical guidelines to make this trustable by other pharmacists.

English

1

0

48

umar@umarelbably·3h

simply, I believe everyone should have access to the best health and medical evidence for free. not just physicians. not sponsored by pharma. patients, biohackers, investors, researchers, healthcare professionals and support functions, builders, students, health conscious. whoever you are, whatever your interest level is — you should have access to honest, transparent, rigorous research. free at frompeer.com

umar@umarelbably

OpenEvidence is a $12B company used by 65%+ of US doctors. I built the version anyone can use for free. Ask Peer a health or medical research question: frompeer.com I thought it was weird that OpenEvidence's model is pharma ads + partnerships with prestige journals; feels dangerous to miss the broader gold standard research. Peer strives for the most honest, transparent, and rigorous answers in health research: - Gold standard sources (PubMed, ClinicalTrials . gov, FDA, NHS, + more) - Source quality weighting - Claim verification (cannot hallucinate citations) - Transparent methodology

English

0

92

umar@umarelbably·4h

so it's a consumer research tool that anyone can use, not a clinical product like openevidence. you still get all the research and evidence on whatever you ask, just not a direct recommendation. auth-gated accounts, your chats are private to your account, no ad pixels on the chat page, full delete anytime. for healthcare professionals we will work towards industry standard privacy protocols.

English

0

46

erik schützler@erikschutzler·11h

@umarelbably how do you ensure privacy?

English

1

0

210

umar@umarelbably·4h

@peptidepedia evidence

English

1

0

3

112

Peptidepedia@peptidepedia·4h

I’m a peptide, scare me with one word

English

37

0

26

2.8K

umar@umarelbably·4h

@0xeevie and “humbled” lol

English

0

1

16

eevie@0xeevie·17h

Why is everyone on LinkedIn so thrilled

English

23

5

111

3.6K

umar@umarelbably·4h

unfortunately incredibly common. i don’t think doctors aren’t turning a blind eye. they do their diligence. patients just don’t feel heard — this is a communication issue — and when that happens they start to look elsewhere.

Viv 🪩@battleangelviv

Tired of not getting taken seriously by doctors, so I’m going to pay for my own full blood panel and whichever else is helpful data and let @grok figure it out

English

0

75

umar@umarelbably·4h

@RenoHemonc also in my day-job I work at Flatiron Health, good to see oncologists engaging in healthtech on here

English

0

10

Santhosh Ambika@RenoHemonc·4h

@umarelbably Great work , kudos !

English

1

0

28

Santhosh Ambika@RenoHemonc·14h

Such low barrier of entry ..

umar@umarelbably

OpenEvidence is a $12B company used by 65%+ of US doctors. I built the version anyone can use for free. Ask Peer a health or medical research question: frompeer.com I thought it was weird that OpenEvidence's model is pharma ads + partnerships with prestige journals; feels dangerous to miss the broader gold standard research. Peer strives for the most honest, transparent, and rigorous answers in health research: - Gold standard sources (PubMed, ClinicalTrials . gov, FDA, NHS, + more) - Source quality weighting - Claim verification (cannot hallucinate citations) - Transparent methodology

English

1

2

401

umar@umarelbably·4h

@OutsourcedLogic @mcuban mark hit me up

English

0

59