Think coffee is a bad habit? Research actually suggests otherwise. This umbrella review of 218 meta-analyses concludes that 3-4 cups a day is associated with lower risk for heart disease and all-cause mortality. Check out the source to read more!
Hey StarTalkians! Season 17, episode 48 was a sit-down discussion with Neil, Chuck and guest Dr. Lika Guhathakurta, a heliophysicist at NASA. The latter half addressed space weather, coronal mass ejections and the many ways in which the sun can – and does – impact our lives.
Super Solar Storms with Lika Guhathakurta - StarTalk Radio
One of Dr. Guhathakurta’s papers delves into what she calls “everyday” space weather, and overlaps with a lot of the podcast discussion.
Carrington-Like Events
Around 50 minutes into the show, they discuss the so-called “Carrington Event” of 1859, and Dr. Guhathakurta also covers this issue in the paper mentioned above. But as her paper points out, there have been others.
A 1989 space weather event knocked out transformers in New Jersey, the United Kingdom and Quebec, with Quebec also losing power for 9 hours. In 2003, geomagnetic storms caused blackouts in southern Sweden and possibly damaged transformers in South Africa.
3 Everyday Effects of Space Weather
Those events are more dramatic – thankfully, we don’t have that type of thing every day. However, that doesn’t mean space weather doesn’t have everyday impacts.
1: Power Problems
It doesn’t have to be a Carrington-like event for space weather to impact your electricity supply. A paper referenced by Dr. Guhathakurta looked at data on insurance claims for industrial electrical equipment, cross-referenced with space weather conditions. While they struggle to pin down a precise dollar amount for the cost – it’s probably in the billions – they calculate that 500 claims are submitted in the US annually because of space weather.
2: Cosmic Rays During Flights
When you travel by plane, you’re exposed to more cosmic rays – high-energy particles originating in distant events like supernovae. At solar maximum, the sun’s magnetic field actually protects us, but at times of lower solar activity, we’re exposed to more. A study looking at this calculated that flying at high latitudes for 12 and a half hours exposes you to as much radiation as a chest X-ray, or about half that for mid-latitudes.
3: The “Blurry Vision” of Your Satellite Navigation
Dr. Guhathakurta estimates [2] that about a third of the inaccuracy of satellite navigation devices come down to space weather. Irregularities in the ionosphere can refract and scatter their signals, leading to what she calls “blurry vision.” It isn’t a huge effect, but it is definitely something you’ll notice often!
Any Infinite Monkey Cage listeners here? Was listening to this episode on Fusion, and they talked about a nuclear fusion reactor in Oxfordshire.
As I understand it, to enable fusion, you need an extremely hot environment.
In the Oxfordshire reactor, scientists conducted an experiment with temperatures so high, that it momentarily became the hottest place in the solar system. Hotter than the sun.
How cool is that. And I found this paper, of a particularly well-known experiment using the reactor (scientists were able to generate a record amount of energy from fusion).
Verdict: False and misleading
“Do you think that the COVID vaccine protected children? [...] Can you show me one study ... that shows that children were protected by the COVID vaccine? [...] that it did more good than harm to children. Can you show me one study that shows that?”
RFK talks vaccines, measles in interview with CNN's Dana Bash – RFK Jr. (from 7:25)
In his recent interview with CNN, RFK Jr. implicitly claimed – via a series of pointed, “gotcha” questions – that there were no studies showing that the COVID-19 vaccine protected children. Explicitly asking about proof “it did more good than harm,” he argued that there is no overall benefit to vaccinating children.
Is he right?
There is Plenty of Evidence that the COVID Vaccine Protects Children
RFK Jr. is absolutely incorrect to imply that no such evidence exists. While most people cannot pull a citation out of their heads on command, that isn’t the same as showing no evidence exists.
A systematic review of such studies was published in JAMA Pediatrics in 2023. It includes data from 15 observational studies and 2 randomized controlled trials focusing on children aged 5 to 11. The total number of participants was almost 11 million vaccinated and over 2.6 million unvaccinated children.
Among other things, this study found a lower risk of any COVID infection (53% reduction), symptomatic COVID infection (47% reduction), hospitalisations due to COVID-related illnesses (68% reduction) and multisystem inflammatory syndrome in children (95% reduction). It also found a 92% increase in the risk of adverse events, but only a non-significant increase in the risk of adverse events that prevent normal daily activities.
What He Might Have Meant: The Risk of Myocarditis
This study did find the well-known risk of myocarditis following vaccination in younger people. However, this amounted to 1.3 cases per million after a first dose and 1.8 per million after a second dose.
Of course, this is not something to discount because of low incidence: it’s a legitimate concern.
However, this risk has to be compared to the risks following COVID infection, which is less likely in vaccinated children. A study from the UK found that there was an increased risk of myocarditis or pericarditis in the first 6 months after vaccination (in young people aged between 5 and 18), but noted that the risk is even higher after COVID infection.
Even without considering other risks (see images), it’s clear that the COVID vaccine is safer than COVID itself, even for myocarditis in young people.
I'm an occasional vegetarian - for health, and environmental and ethical reasons. I think most would already agree that one can have a complete diet, without meat. Nonetheles, I found this paper useful.
It's a narrative review of evidence. A few helpful takeaways (gleaned from the Credo summary):
A plant-based diet has health benefits, and is better environmentally.
A plant-based diet can be complete, but there are some potential drawbacks. E.g. it's harder for the body to absorb certain nutrients from plant sources.
And so, if you're a vegetarian or vegan, you do have to put in some thought, to ensure your diet is complete. (Though the same could be said of any diet!)
I had known that plants generally "breathe" and release carbon dioxide at night. But I didn't know that plants also "fart".
Learnt about this listening to this Brains On episode with my kid!
And here's the paper that published the discovery! The findings (from credo paper summary):
The study reveals that many types of terrestrial plants can produce methane under normal oxygen-rich conditions, which was not previously recognized. The authors demonstrate that this methane is formed inside the plant tissues, not by microbes in the soil.
So plants produce methane "farts" in a sense!
(The Brains On hosts were quick to say that despite this, trees/forests are generally carbon sinks i.e. they do not cause global warming 😀)
A new study has found a new insight on the relationship between blood pressure and cognitive decline. Since hypertension is a well-established – and controllable – risk factor for cognitive decline, scientists assumed that reducing it is the best strategy. But this seems to be missing a crucial piece of the puzzle.
The Problem: Mixed Results on Lowering Blood Pressure
The new paper points out that different studies disagree about the cognitive benefits of interventions to reduce blood pressure.
The SPRINT-MIND study recruited people with hypertension but not diabetes. It found a reduced risk of mild cognitive impairment for people on a more intensive blood pressure reduction regimen, but no difference in probable dementia.
Conversely, the ACCORD-MIND trial focused on people with type 2 diabetes and included a similar comparison of a more intensive vs. less intensive blood pressure reduction regimen. In this case, though, there was no difference in cognitive function.
What the New Study Did
The new research [1] pooled data from both of these trials, and analyzed the data again at participant-level. The paper notes that blood pressure variability (BPV) is linked to both cardiovascular events and cognitive decline, and tends to be higher in people with diabetes.
To test whether this is the important factor, researchers looked at BPV between each visit in the original studies. The original researchers used varying cognitive tests, but the authors identified two essentially equivalent measures of psychomotor processing speed among them. They used these to compare results across both studies.
What They Found: Higher Variability, Greater Decline
People with more variability in their blood pressure saw greater cognitive decline over the study period. In particular, those in the highest third of BPV had a faster rate of decline than those in lower thirds. The relationship seemed to be linear: each 10% increment of BPV led to consistently higher levels of decline.
What the Study is Missing
The study only focused on one measure of cognitive decline, rather than a global score. All participants also had a high cardiovascular risk, so the results might not carry over to other populations.
What Doctors Can Do
One thing the study suggests is optimizing drug selection and timing to reduce BPV where possible. Doctors can also try novel interventions specifically designed to bring down BPV.
For years, AdamW has been the undisputed workhorse of large language model (LLM) pretraining. 🖥️
AdamW is an element-wise, first-order optimizer that updates every parameter independently: simple, reliable and easy to scale. It powers many frontier models today. However, as researchers push toward larger models, longer training runs, and massive batch sizes, AdamW is beginning to show its weaknesses.
NVIDIA researchers found at global batch sizes approaching 100 million tokens per step, AdamW begins to lose effectiveness. hence a new study from NVIDIA explores a fundamental question:
Can we build optimizers that are smarter than AdamW and allow future AI models to train faster, more efficiently, and more reliably?
Unlike AdamW, which treats parameters independently, Muon and SOAP introduce structure-aware optimization, methods that use information about the geometry and relationships within neural network weights.
Matrix-aware updates: Optimizes the geometry of gradient updates rather than scaling parameters independently like AdamW.
Balanced learning directions: Uses Newton-Schulz iterations to orthogonalize momentum updates and balance singular values.
Memory efficient: Removes the need for AdamW-style second-moment statistics, reducing optimizer memory.
Better scaling: Maintains stable training at larger batch sizes and model scales.
Uses parameter structure: Applies Shampoo-inspired Kronecker preconditioning to capture row and column correlations in weight matrices.
Better optimization space: Rotates gradients into an eigenbasis where directions become far more effective to update.
Adam-compatible: Performs adaptive updates like AdamW but in a more informative coordinate system.
Second-order benefits: Gains curvature awareness without the prohibitive cost of full second-order optimization.
📌 This shift towards new structure-aware optimization approaches would allow for:
More efficient AI development
Lower computational costs
More capable future models
👉 To make this transition possible, NVIDIA has released open-sourced implementations in Megatron-LM and a standalone Emerging-Optimizers library to be available to the public.
AdamW is unlikely to disappear and will remain a staple for small-to-medium training runs. Its simplicity and maturity make it an excellent choice for many applications.
╰┈➤ˎˊ˗ But for frontier-scale training? It still remains to be seen, where it'll go next.
What if a blood sample could reveal which parts of your body are aging the fastest, and even predict your risk of developing diseases years before symptoms appear? 🤔❓
A new study published by Stanford researchers analyzed blood plasma proteins with nearly 3,000 proteins across 44,498 UK Biobank participants for organ aging, and 7,000+ proteins in 60,542 individuals for cell-type-specific aging.
Using Machine Learning, the scientists figured out how much composite protein “signature” represented each organ and let the models assign biological ages to 11 organ systems and 40+ cell types.
😮 Turns out one-third of us have at least one organ that's aging dramatically faster or slower than the rest.
They discovered that aging is far from uniform. Instead, different cells and organs age at different rates, meaning two people of the same chronological age can have very different biological aging profiles, where 1 in 4 participants had multiple extremely aged or youthful organs; 20–25% showed accelerated aging in a single cell type, while 1–3% had >10 rapidly aging cell types.
In the study, they claim that cellular aging signatures were uniquely associated with disease status:
🧠Astrocytes (brain support cells): Extreme astrocyte aging demonstrated a 12.59-fold increased risk of incident Alzheimer’s disease (AD).
💪🏻Skeletal muscle cells: People with aged muscle cells had a 12.7× higher risk of amyotrophic lateral sclerosis (ALS), detectable three years before symptoms appear.
🫁 Respiratory cells: Smokers with aged lung epithelial cells faced 58% higher lung cancer risk than smoking alone would predict.
🍀Interestingly, new findings on Alzheimer’s disease offer hopeful news for carriers of the APOE4 gene that has long been linked to Alzheimer’s.
The predictive power of astrocyte aging (HR = 5.16) was comparable to APOE4 carrier status itself (HR = 5.30).
APOE4 gene + aged astrocytes three times higher risk compared to an average APOE carrier.
In contrast, APOE4 carriers with youthful astrocytes appeared to neutralise the genetic risk completely as none of their 23 APOE4/4 carriers developed Alzheimer's.
Youthful astrocytes reduced overall AD risk by over 60%
Side Note: ♀️ Women seemed to be more vulnerable to the harmful associations in both APOE4 gene and aged astrocyes.
🧬The researchers also identified an inverse pattern via biological age quantification from blood for the APOE gene variants:
APOE4 → astrocytes age faster, macrophages age slower
APOE2 (the protective variant) → the exact opposite: younger astrocytes, older macrophages
Together, these findings highlight a shift in how we understand aging. Chronological age is merely one measure, and this algorithm may provide a more precise picture of one's health trajectories by decoding cellular aging signatures from blood.
✨ Check out the these links for a deeper read:
More on the APOE Gene 🧬
Stanford Medicine Newsletter: Biological age of our organs and cell types within them predict our disease risk, longevity.
It was recently reported that Meta was ordered to pay $567M in additional fines for an abatement program in New Mexico due to its failure to warn the public about its platforms' risks to children. The fines were in addition to $375M ordered against Meta for a similar earlier claim in New Mexico. This marks the largest child safety ruling against Meta and will have reverberations for social media companies in the future.
The court case ruled Meta acted as a public nuisance infringing on the health and safety of the public "knowingly" and "without lawful authority". The state built its case around these 10 components:
New Mexico is experiencing a youth mental health crisis.
Problematic social media use is a cause of the youth mental health crisis.
Meta’s platforms are widely used in New Mexico.
Meta implemented platform features that were designed to maximize engagement and time spent on its platforms, for all users and for teenage users in particular.
Meta’s engagement features encourage and facilitate problematic interactions between adults and children.
Harm caused by Meta’s products affects a significant number of people in New Mexico.
Although many social media platforms compete for the attention of New Mexico’s youth, Meta’s platforms rely on methods for optimizing engagement that disproportionately impact education and interpersonal relationships
Meta and its platforms provide a valuable service in New Mexico despite these harms, and it is not the only social media company impacting youth.
Meta knowingly designed its platforms to maximize engagement, including in ways that are harmful to teenagers, and does not adequately disclose the risks of harm to its users.
Meta’s conduct is a cause of and significantly contributes to the youth mental health crisis in New Mexico.
Deep dive into evidence of harm towards youth
The court case itself didn't provide any specific papers as far as I can tell, only references to HHS advisories, reports, and internal Meta research. The evidence was built on expert testimony, which the court found strong enough to state:
Even without the evidence and testimony that certain percentages of youth ... would not have those conditions but for social media use, the highly probable result of Meta's engagement optimization practices ... is that such practices would increase anxiety, depression, and sleep deprivation among teenagers.
The experts on both sides provided counter testimony on their view of the connection between social media and harm to youth. Dr. Jean Twenge testified that
most longitudinal studies confirm that adolescents’ social media use leads to depression ... and that experimental studies show that people who eliminate or reduce social media are less lonely and depressed, which further confirms the causal path between social media and teen mental health harms.
While the evidence wasn't listed, a 2022 study Social Media and Mental Health backs this testimony showing that the arrival of Facebook on college campuses lead to worse student mental health. The HHS advisory on Social Media and Youth Mental Health also listed this 2019 JAMA Psychiatry study.
Meta on the other hand pointed to a 2024 consensus report on Social Media and Adolescent Health which pointed to other factors causing overall mental health decline and a complex role social media plays in society.
My take
There does seem to a slew of evidence pointing to correlation between social media and declining mental health with teens and adolescents. Additionally, the ability for online predators to connect to underage minors is also blatant. Arturo Bejar, who worked as Meta’s Senior Director of Engineering in charge of the Protect and Care team stated:
the product is very good at connecting people with interests, and if your interest is little girls, it will be really good at connecting you with little girls.
But I agree with the primary conclusion from the Social Media and Adolescent Health report. There are benefits that are associated with social media including connection to friends and combating isolation, civic engagement, learning, and self expression.
The issue arises when we try to simplify the debate to net benefit or harm. Instead, its better to consider the piecemeal impacts and combat clear harms while retaining clear benefits. I don't think its helpful to see social media as a pure harm, just the same I wouldn't categorize screens and screentime as a pure harm.
