Fabian Wahle

Counter-Evidence Studies • CTT Collaboration (2010): The Lancet (26 RCTs, 170,000 participants)
Key takeaway: More intensive statin therapy produces further proportional reductions in major vascular events (15% extra per 0.51 mmol/L additional LDL drop) and all-cause mortality (10% per 1 mmol/L LDL reduction), with a clear linear dose-response and no lower threshold. thelancet.com • CTT Collaboration (2012): The Lancet (27 RCTs, ~174,000 participants)
Key takeaway: Statins reduce major vascular events by 21% per 1 mmol/L LDL-C drop even in low-risk people (<10% 5-year risk), yielding absolute benefits (11 fewer events per 1,000 over 5 years) that greatly exceed known harms—benefits consistent across age, sex, and baseline LDL. thelancet.com • Silverman et al. (2016): JAMA (49 trials, 312,175 participants)
Key takeaway: Every 1 mmol/L LDL-C reduction (via statins or non-statins that upregulate LDL receptors) cuts major vascular events by 23% (RR 0.77); benefits are proportional to the LDL drop achieved, and lower attained LDL levels directly correlate with fewer coronary events. pubmed.ncbi.nlm.nih.gov • Navarese et al. (2018): JAMA (34 trials, 270,288 participants)
Key takeaway: Intensive LDL-C lowering reduces all-cause mortality (RR 0.92) and CV mortality (RR 0.84), with progressively greater benefits at higher baseline LDL-C (>100 mg/dL); no mortality gain when baseline LDL is below 100 mg/dL. jamanetwork.com • Burger et al. (2024): Atherosclerosis (60 RCTs, 408,959 participants)
Key takeaway: LDL-C reduction yields a consistent HR of 0.78 for major vascular events per 1 mmol/L drop; benefits remain stable over time (up to ~10 years) in secondary prevention and do not meaningfully attenuate except slightly with older age in primary prevention. atherosclerosis-journal.com • Peters et al. (2016): Atherosclerosis (97 cohorts, >1 million people)
Key takeaway: Raised total cholesterol is a strong independent risk factor for coronary heart disease in both sexes (RR 1.20 in women, 1.24 in men per 1 mmol/L increase), with only a tiny male advantage; it has essentially no effect on total stroke risk. pubmed.ncbi.nlm.nih.gov • Ridker et al. (2024): New England Journal of Medicine (Women’s Health Study, 27,939 women, 30-year follow-up)
Key takeaway: Higher baseline LDL-C strongly predicts 30-year major CV events in women (top vs. bottom quintile HR 1.36); LDL-C, hs-CRP, and Lp(a) each contribute independently, and combining all three biomarkers gives the strongest long-term risk prediction. pubmed.ncbi.nlm.nih.gov

My doctor put me on a statin after my heart attack at 52. I trusted him. I took the pill. I never asked a question. Then I found this study. 60 clinical trials. 323,950 people. Every cholesterol lowering drug ever made. Statins. PCSK9 inhibitors. Ezetimibe. They measured how much each drug lowered LDL cholesterol. Then they measured whether people lived or died. The line is flat. It did not matter if they lowered LDL by 10% or by 70%. The death rate did not change. In some trials people died more. 323,950 people. Near zero benefit. Published in the Journal of Cardiovascular Pharmacology. 2023. Nobody showed me this chart. Not my cardiologist. Not my pharmacist. Not the drug rep who visited my doctors office every month. I had to find it myself. After the heart attack.

@cosmicsurfer6 @PaulAustin3w AA = active-active PA = placebo-active Both groups receive 2 doses, where AA receives 2x DMT and the other group receives 1x placebo and 1x DMT. Note: Due to dropout, the last data point is not statistically significant and can be 'random'

@PaulAustin3w Paul, can you help me understand this? PA = psilocybin and AA = nn-DMT. I am curious about dosages. It would appear the PA group saw more gains post 2nd dose. As if it was finishing off a job. If the initial PA dose was 150% of original then what? Still, nine week rule

We may be entering the “psychedelic lunch break” era. New clinical data just dropped on 10-minute IV nn-DMT infusions, w/ rapid antidepressant effects lasting months. Same efficacy as psilocybin but in a fraction of the time. Most importantly, no aliens were harmed in the process.

1. Triple-Negative Breast Cancer (TNBC) This is currently where the strongest evidence for improved survival exists. • The Study: A sub-analysis of a clinical trial (NCT03340935) published in the International Journal of Cancer and Cell Metabolism investigated cyclic fasting-mimicking diets (FMD) with first-line chemotherapy.  • Results: Patients with advanced triple-negative breast cancer who followed the FMD had a median overall survival of 30.3 months, compared to 17.2 months in those on a regular diet.  • Other Findings: The DIRECT-1 trial also showed that a 5-day FMD significantly increased the likelihood of a "pathological complete response" (the disappearance of all signs of cancer in tissue samples) after chemotherapy. 2. Colorectal Cancer Colorectal tumors are highly sensitive to glucose levels, making them a primary target for metabolic interventions like fasting.  • The Findings: Clinical trials (such as NCT04247464) have shown that fasting for 48 hours around chemotherapy cycles enhances the immune response against the tumor and reduces toxicity.  • Survival Impact: Preliminary data and reviews suggest that fasting may improve survival outcomes by approximately 20% in colorectal patients by activating autophagy and reducing tumor-promoting inflammation.  3. Lung Cancer (NSCLC) Fasting appears to work best for lung cancer when combined with either chemotherapy or newer immunotherapy drugs. • Mechanism: Studies indicate that fasting inhibits the IGF-1 axis, which helps "unmask" lung cancer cells to the immune system.  • Results: Research published in Nature suggested that 24–48 hour fasting cycles can substantially improve the body's response to checkpoint inhibitors (like PD-1/PD-L1 blockers), potentially extending survival in patients who might otherwise be resistant to treatment.

If I had cancer, I would do this immediately…

@apstrusus84 @cremieuxrecueil Familiarize yourself with how small interfering RNA mechanisms work (there are fantastic YouTube videos available) — it may already help with your concerns.

@cremieuxrecueil Any research you could point me to that would help me become more comfortable with that as a delivery method? I obviously have a need for it, but my hesitance I believe is somewhat warranted, from the limited research I've reviewed. 2/2

It's happening! Lipoprotein(a) testing is going to skyrocket once it's possible to treat it.

Neat! Could you run your ruleset on a bench completely independent from SWE-Bench and compare the performance? Hunch here is that you fell for inherited bias the way you used your optimisation loop i.e albeit splitting you may have overfitted on SWE-Bench set having an outer loop. Keep diggin! :)

We improved @cline, a popular open-source coding agent, by +15% accuracy on SWE-Bench — without retraining LLMs, changing tools, or modifying Cline's architecture. We achieved this simply through optimizing its ruleset, in ./clinerules — a user defined section for developers to add custom instructions to the system prompt, just like .cursor/rules in Cursor or CLAUDE.md in Claude Code. Using our algorithm, Prompt Learning, we automatically refined these rules across a feedback loop powered by GPT-5. Here’s how we brought GPT-4.1’s performance on SWE-Bench Lite to near state-of-the-art levels — matching Claude Sonnet 4-5 — purely through ruleset optimization. See our more detailed blog post 👉: arize.com/blog/optimizin…

@wonderofscience meh.. healthcare-in-europe.com/en/news/11-7-t…

The highest-resolution MRI scan of a human brain ever, capturing details as small as 100-microns, achieved with a 7-tesla magnet and custom software by researchers at Massachusetts General Hospital.

What it is. A new architecture called the Hierarchical Reasoning Model (HRM) that aims to do multi-step reasoning inside the network’s hidden states rather than by emitting long chain-of-thought text. It was posted to arXiv on June 26, 2025 and revised July 22, 2025.  What’s new. •Two coupled recurrent modules run at different speeds: a slow, high-level “planner” and a fast, low-level “executor.” This lets the model plan and then refine, multiple times, within a single forward pass.  •A training trick approximates backprop through time with a one-step gradient, keeping memory constant and training stable—no full BPTT unroll.  •Emphasis on “latent reasoning” (thinking in hidden states) instead of producing explicit step-by-step text.  How it works (intuitively). The fast module iterates to settle on a local solution; then the slow module updates the plan and “resets” the fast module to work from that new context. Repeat. This hierarchical convergence yields many effective computation steps without making the network extremely deep.  Results they report. •With only ~27M parameters and ~1,000 training examples, HRM nearly solves hard Sudoku and 30×30 maze tasks where CoT methods failed.  •On ARC-AGI, trained from scratch (no pretraining, no CoT), HRM reports ~40.3%, surpassing larger CoT-based models (e.g., o3-mini-high at 34.5%, Claude 3.7 8K at 21.2%).  Why it matters. It’s a concrete step toward reasoning-centric, recurrent systems that compute deeply without token sprawl—an alternative to “just scale the Transformer + longer CoT.”  Caveats. This is a fresh preprint; results are on specific benchmarks and need independent replication and broader tests (e.g., open-ended tasks). Also, despite the “no pretraining” claim, performance in the wild may differ from curated puzzles.  

@makingAGI can we get a breakdown thread on this?? feels too good not to explain further

🚀Introducing Hierarchical Reasoning Model🧠🤖 Inspired by brain's hierarchical processing, HRM delivers unprecedented reasoning power on complex tasks like ARC-AGI and expert-level Sudoku using just 1k examples, no pretraining or CoT! Unlock next AI breakthrough with neuroscience. 🌟 📄Paper: arxiv.org/abs/2506.21734 💻Code: github.com/sapientinc/HRM

apologies if already suggested, lots of low hanging fruits already mentioned, but try Sleep Restriction Therapy. If that doesnt work try some well researched mental health scores (eg phq9 or becks index for depression). depending on your burden- seek advise from a sleep professional / clinique. Good night!

i am SICK of this. please share anything that has helped you sleep better, even the weird stuff. i am desperate and i’ll try anything!!!! (this is a cry for help)

@papipnl There won't be 180 day 1:1 redemption tho. Take a closer look what exit option 1 is about (need to burn USUAL)

There's been a lot of panic going around but I just locked in some nice yield on the June26th USD0++ PT. The math: Best Case (1:1 usd0++ to usd0 redemption): 17.716% * (365/166) = 39% APY Worst case (180 days for 1:1 redemption): 17.716% * (365/346) = 18.7% APY

The debate over the so-called “lack of transitional fossils” is increasingly outdated, especially in light of the monumental advances in DNA sequencing and genetic research over the past few decades. The argument that we don’t observe transitions between species ignores the overwhelming genetic and molecular evidence that fills any perceived gaps in the fossil record. Genetic Evidence Bridges the Gaps Modern DNA sequencing technologies have allowed scientists to compare the genomes of different species with unprecedented precision. These comparisons reveal a tapestry of genetic similarities and differences that map perfectly onto evolutionary relationships predicted by the theory of evolution. For example: • Endogenous Retroviruses (ERVs): These are viral sequences embedded in the genomes of organisms. The pattern of shared ERVs among species mirrors the evolutionary tree, providing compelling evidence of common ancestry. • Genetic Mutations and Natural Selection: We can directly observe mutations occurring in populations over time. Studies of organisms with short generation times, like bacteria and viruses, have documented speciation events and adaptations in real-time. • Molecular Clocks: By measuring the rate of genetic mutations, scientists can estimate the time since two species diverged from a common ancestor, corroborating fossil and morphological data. Transitional Fossils Do Exist While it’s true that the fossil record is incomplete—owing to the rare conditions required for fossilization—we have discovered numerous transitional fossils that showcase the gradual changes between major groups: • Tiktaalik roseae: Fills the gap between fish and early amphibians, exhibiting features of both aquatic and terrestrial animals. • Archaeopteryx: Displays both avian and reptilian characteristics, bridging dinosaurs and modern birds. • Australopithecus afarensis (e.g., “Lucy”): An early hominin that shows traits intermediate between modern humans and our primate ancestors. These fossils are snapshots of evolutionary transitions, providing tangible evidence that counters the claim of missing links. Every Species is Transitional The concept of distinct “species” is a human classification tool. In reality, evolution is a continuous process with populations constantly undergoing genetic changes. Just as we don’t notice day-to-day changes in a growing child but recognize the difference over years, evolutionary changes accumulate over generations: • Ring Species: Certain populations can interbreed with neighboring populations, but the ends of the range cannot interbreed, illustrating gradual change over geography. • Hybrid Zones: Where ranges of different species overlap, hybrids can form, demonstrating ongoing genetic exchange and evolution. Observable Speciation Events Speciation is not just a historical occurrence but an observable phenomenon: • London Underground Mosquito (Culex pipiens f. molestus): A new species that adapted to the underground environment, reproductively isolated from its above-ground counterparts. • Darwin’s Finches: Ongoing studies show how environmental pressures lead to changes in beak shapes and sizes, a precursor to speciation. Evolutionary Developmental Biology (Evo-Devo) Advances in evo-devo have unveiled how small genetic changes can lead to significant morphological differences: • Hox Genes: These genes control the body plan of an embryo. Mutations here can result in significant anatomical changes, providing a mechanism for major evolutionary transitions. • Gene Duplication: This process can introduce new genetic material for evolution to act upon, leading to new functions and traits. So.. The assertion that we lack transitional fossils or observable evidence of species transitioning is not supported by the wealth of data available. Evolutionary theory is robust, backed by interdisciplinary evidence from genetics, paleontology, molecular biology, and developmental biology. The supposed “missing links” are a misinterpretation of how gradual change and classification work. As our technologies and methods improve, the evidence for evolution continues to grow, rendering the lack of transitional fossils an argument of the past.

@bsilone @waitbutwhy These are analogies to help you understand it, not arguments. The evidence is overwhelming, but way too much for X. If you want to learn, you can find it yourself, but you clearly don’t want to, so I’ll let you be.

Yes, because sharks and fish branched off before fish diversified into whales. So all mammals, reptiles, birds, and other fish and more closely related to each other than any of us are to sharks. We're the equivalent of siblings and sharks are our cousins.

@foundmyfitness @YouTubeCreators Hang in there Rhonda. You got this!

I’ve received hundreds of angry comments precisely because I’ve shared fundamentally PRO-vaccine positions. In spite of that, I’m getting slammed by @YouTubeCreators takedowns. Any analysis is now taboo because, to an unsophisticated moderator, it all sounds the same.

Nash is integrating with Polygon (@0xPolygon), formerly known as Matic! Polygon is a scaling solution for #Ethereum that is being adopted widely. Nash will list the $MATIC token and integrate our fiat gateway with the Polygon network. Read more! 👇 1/4 blog.nash.io/profile-polygo…

This afternoon the FMA of Liechtenstein added Neon Exchange AG – Nash's parent company – to its register of regulated exchange bureaus: ID 261096, fmaregister.fma-li.li/search?searchT… Nash Cash will launch on 14 September. Information on Nash Pay will follow. #Nash #TrustYourselves $NEX

@ETFCorp @canesin reach out on telegram @fabwa

@fabwa I tried sending you a DM but it didn’t go through, so I will try it this way. A while back I spoke to @canesin on telegram and he advised me to approach you. We are a startup currently developing a product, and we would like to discuss legal details with you personally.

@elastosinsights @elastos_sash still hearing those 🦗 💚

@elastos_sash 2017 .... Bought "antshares" NEO for 17c..There was only a few guys in the slack some scammers and some crickets... One of the guys went on to create @nashsocial ...@fabwa

When did you first get into crypto? Any memories / stories from the early days?

Nash recently made our client protocol and SDK public on GitLab. Our open-source ethos shows our commitment to transparency and innovation around the Nash platform. Read more about our strategy here! #Nash #TrustYourselves blog.nash.io/nash-is-open-s…

The first version of our Bitcoin trading protocol with secure multi-party computation (MPC) API keys is now open for public testing on app.sandbox.nash.io. For security vulnerabilities we reward responsible disclosure at bugbounty@nash.io.

@UltimateCrypto7 @neo_block @tezos its @NEO_Blockchain ...

Which is the better crypto? @NEO_block @tezos

USDC markets are now available in Europe! Nash users in European jurisdictions can enjoy fully compliant trading against a stablecoin. #Nash #TrustYourselves $USDC

Great covarage of #nash @nashsocial published today on #forbes @forbes @canesin @fabwa @unignorant $NEX $BTC $ETH forbes.com/sites/ktorpey/…