Saloni

@scientificdiscovery

Co-founder & editor, Works in Progress. Writer, Scientific Discovery. Podcaster, Hard Drugs. Advisor, Coefficient Giving. // Previously at Our World in Data. Newsletter: Podcast: 🏳️‍🌈

1,172 Following44,888 Followers5,102 Posts

YouTube YouTube Website Website Bluesky

education podcasts science

Is this your profile?Claim it and scientificdiscovery.dirty.social is yours.Claim profile

Saloni@scientificdiscovery· 27d

Sure you've probably seen bored pooh and maybe fancy pooh. Possibly you've even seen royal pooh. But have you seen medieval pooh?

Saloni@scientificdiscovery· 28d

Advantages of jointly tackling trial and market design

What are the major advantages of my suggestion?

Because everything is done in the same trial, it is cheaper, faster, and statistically more efficient (less bias, more power).

A shared design allows the funder to make better head-to-head comparisons between these antivirals. In some cases it also avoids companies from competing for trial subjects.

It can de-risk a sensitive part of the R&D process for firms. For producers many uncertainties revolve around whether their intervention can feasibly be tested. Under this proposal the funder gets control of the information that is generated, but assumes (some of) the burden of creating the testing infrastructure. This solution can also be thought about as an intermediate solution between push and pull funding—especially if the funder also covers some/all costs of the trial.

Saloni@scientificdiscovery· Jul 12

Went to the postal museum and rode the mail rail! I'm not sure how to feel about the fact that I'm small enough that I didn't have to bend or crouch inside it at all.

Saloni@scientificdiscovery· Jul 12

The video is also part of a fundraising campaign by Giving What We Can to help the effort to eradicate malaria. They've raised $1.7M from it since last week! givingwhatwecan.org

Saloni@scientificdiscovery· Jul 12

Malaria elimination succeeded in much of the world, including Europe, North America, Australia, and much of Asia…

Malaria prevalence over the twentieth century - map of regions where malaria was eliminated.

Saloni@scientificdiscovery· Jul 12

I've lived here since 2013 but learnt for the first time today that there's a Benjamin Franklin museum in London

Card room. Some cards laid out on a table with Benjamin Franklin memorabilia behind them

Saloni@scientificdiscovery· Jul 11

Great post from @drjenndowd.bsky.social about what's wrong with these 'turning point' in aging studies jenndowd.substack.com/p/does-adole... with a great example by @adamjkucharski.bsky.social:

Does the brain really change at four major turning points around ages nine, 32, 66 and 83?

Histogram of participants in the dataset by age, and smoothed average

Saloni@scientificdiscovery· Jul 11

I don't know how people can look at this and think to themselves "that seems plausible" (rather than e.g., maybe this is an artefact of participation into the samples) Demographers, have you ever seen a health trend over age that looks like this?

A line chart showing the rate of aging according to 'biological clock' measures, except it looks like a roller coaster.

Saloni@scientificdiscovery· Jul 11

Not again!

Peter Gray not realising that the survey methodology had changed between 2009 and 2013, as well as the question phrasing.

Saloni@scientificdiscovery· Jul 9

Saloni@scientificdiscovery· Jul 7

Darn. How will I get answers to the important questions in my life?

me: if siblings share 50% of their DNA with each other, and humans share 50% of their genes with bananas, is my brother as related to me as a banana?

claude fable: *switched to Opus 4.8*

Saloni@scientificdiscovery· Jul 7

A question far too dangerous for Fable to answer.

me: why did the chicken cross the road?

claude fable: *switched to Opus 4.8*

Saloni@scientificdiscovery· Jul 5

Found a Gizmodo article about it gizmodo.com/researchers-... which says they could be used for “search and rescue missions” — just like I suspected!

Saloni@scientificdiscovery· Jul 5

Just started listening - I've been looking for a book on this topic for years and only just found that it exists!

To End A Plague. America's fight to defeat AIDS in Africa. By Emily Bass.

Randy Shilts and Laurie Garrett told the story of the HIV/AIDS epidemic through the late 1980s and the early 1990s, respectively. Now journalist-historian-activist Emily Bass tells the story of US engagement in HIV/AIDS control in sub-Saharan Africa. There is far to go on the path, but Bass tells us how far we’ve come.” —Sten H. Vermund, professor and dean, Yale School of Public Health

With his 2003 announcement of a program known as PEPFAR, George W. Bush launched an astonishingly successful American war against a global pandemic. PEPFAR played a key role in slashing HIV cases and AIDS deaths in sub-Saharan Africa, leading to the brink of epidemic control. Resilient in the face of flatlined funding and political headwinds, PEPFAR is America’s singular example of how to fight long-term plague—and win.

To End a Plague is not merely the definitive history of this extraordinary program; it traces the lives of the activists who first impelled President Bush to take action, and later sought to prevent AIDS deaths at the whims of American politics. Moving from raucous street protests to the marbled halls of Washington and the clinics and homes where Ugandan people living with HIV fight to survive, it reveals an America that was once capable of real and meaningful change—and illuminates imperatives for future pandemic wars. Exhaustively researched and vividly written, this is the true story of an American moonshot.

Saloni@scientificdiscovery· Jul 3

Apparently the Gila monster is pronounced 'HEE-la' - there's no G sound - which I find especially confusing in a biology context because hearing 'Gila monster' makes me think of HeLa cells gone rogue.

Saloni@scientificdiscovery· Jul 2

This piece on peptides is very interesting on the whole. But most interesting to me was this brief history of how wearable closed-loop insulin systems (i.e. both monitoring blood glucose and releasing insulin accordingly) were developed by biohackers before they were developed commercially.

“The risk was in not doing it.”

Dana Lewis was blessed with a problem doctors recognized and knew how to treat, more or less. Her body didn’t produce enough insulin. The treatment for this condition is straightforward — more insulin — but it requires exquisitely precise dosing. If a diabetic injects too much insulin, their blood sugar will crash, which, left unchecked, can be fatal. Administering too little risks high blood sugar, which slowly damages organs.3

Today, diabetics can monitor their blood sugar and choose their insulin dose with wearable continuous glucose monitors, which alert them when it falls or spikes to dangerous levels. But back in 2014, when Dana’s story started, continuous glucose monitors (CGMs) were new to the consumer market and not yet considered reliable. Their function was to trigger a blood stick check or serve as a last-minute alert to catastrophically low blood sugar.

But these monitors aren’t perfectly reliable, and they were even less so in 2014. The sensor in Dana’s could be thrown into a panic just by lying on the arm it was attached to, so false alarms were common. After numerous false alarms, she learned to sleep through the alerts. She complained to the manufacturer, who told her, “It works for most people” — as if that solved her problem.

Dana’s biggest fear was having a hypoglycemic episode in her sleep and dying before she had a chance to do something about it. She did not find the fact that the app woke up other people in time very comforting. So she and her new boyfriend (now husband) Scott Leibrand set to work on a solution.

First, they needed real-time access to the data from the CGM, the only way to identify problems as they happened. Luckily John Costik, a software engineer and parent of a diabetic preschooler, had already created Nightscout, a program to extract data from CGMs and upload it to cloud storage. Scott, also a software engineer, wrote a program to monitor these uploads, and if they reported low bloo…

This solved the immediate problem of Dana sleeping through her own death but still left her with the problem of managing diabetes during the day. This process was time-consuming at best, and many people didn’t manage it well, spending long periods of time with their blood sugar damagingly high or dangerously low.

In addition to continuous glucose monitors, state-of-the-art diabetic treatment included an insulin pump. These still required users to calculate their dose based on a mix of scientific literature and their personal history, and manually program the device, but allowed for more complicated dosing schedules over longer periods of time, instead of a single large bolus from a needle.

So Dana, Scott, and other collaborators set out on a bigger project to “close the loop.” They wrote an algorithm that pulled the data from the CGM, applied the same math patients did by hand, and fed it to a hacked insulin pump.4 This created much more responsive, sophisticated dosing that gave them more time in the ideal blood sugar range and less risk of catastrophic failure while requiring less attention.

One reason Dana felt safe writing her own pancreas was that she could do so in very small steps. OpenAPS programmers erred on the side of undersupplying insulin, but the user could get more aggressive over time. OpenAPS also respected the built-in limits of the insulin pump, which meant the system would have to work very hard to do more harm than a human could do from inattention.

When I asked how she thought about the risk of writing her own pancreas, Dana said that all of her risk was in accepting the status quo. She knew a pharma company would release their own closed-loop system eventually — the idea was too obvious and too valuable not to — but she was unwilling to risk dying in the meantime.

Saloni@scientificdiscovery· Jul 1

Here, have a potion

Saloni@scientificdiscovery· Jul 1

Oh no! Oh no!!

In most cases, the more covariates we have in the data, the higher the chances we catch all confounders. (Overfitting can be a risk in prediction problems, but in public health there are usually too few rather than too many covariates in the model.) This is why, in a world with a growing wealth of data, observational evidence treated with causal methods ends up winning.

Saloni@scientificdiscovery· Jun 30

This piece features some of my favourite dataviz, made by our new specialist, Manuel Bortoletti!

Wegener's puzzle: Glaciers once traversed continents that are now tropical. Map of continents and glacier movement across them.

Wegener's puzzle: Similar fossils are found across continents. A map illustration of ancient species across continents.

Diagram: The direction of magnetic north wanders and occasionally reverses.

Saloni@scientificdiscovery· Jun 28

Some people looking at a README file:

That sign can't stop me because I can't read!

Saloni@scientificdiscovery· Jun 28

Many forget this, and we all suffer for it.

You wouldn't download a dataset without reading the data quality notes

Saloni@scientificdiscovery· Jun 22

Most people don't realise how lovely it is to travel through London by ferry, on the Thames Clippers/Uber boat! It goes pretty fast in outer London, and there's a bar on board 😂

Saloni@scientificdiscovery· Jun 20

My nephew watched Toy Story 5 today I asked him which one was his favourite character and he pointed to the iPad

Saloni@scientificdiscovery· Jun 19

This is cool - I'm in the weights of several LLMs. intheweights.com But I find it strange that most LLMs confidently assert that I'm British (I'm not!)

My result on In The Weights dot Com. I am in the weights for AI models, in the top 10%.

Saloni@scientificdiscovery· Jun 19

Very interesting talk by @mendelrandom.bsky.social on the apparent effect of shingles vaccines on dementia. Clear discontinuity from shingles vaccines on shingles, but not on dementia, in this much larger follow up in England, using similar analytic choices as the Wales study that showed a benefit.

RDD of cumulative risk of shingles over time, with a sudden drop when shingles vaccines are introduced.

RDD of cumulative risk of dementia over time, with no clear discontinuity when shingles vaccines are introduced.

Saloni@scientificdiscovery· Jun 17

Should I buy this incomprehensible t-shirt?

I have no Purpose and No
in a world
destination
I'm just
Wondering arou
World where
Can't possess anything

Saloni@scientificdiscovery· Jun 16

Everything looked a LOT smaller than before though.

Saloni@scientificdiscovery· Jun 15

I think we all know which platform teenagers will return to once Instagram, Snapchat, Reddit, TikTok, YouTube, Reddit Twitch, Threads, and X are banned

Saloni@scientificdiscovery· Jun 15

I think we all know which platform teenagers will return to once Instagram, Snapchat, Reddit, TikTok, YouTube, Reddit Twitch, Threads, and X are banned

Saloni@scientificdiscovery· Jun 14

Little bugbear when people describe EGFR+ lung cancers as 'a rare type of lung cancer'. While it is a minority share of lung cancers, lung cancer is the most common cancer by far, so EGFR+ lung cancers are a common cancer. Its annual deaths (~16k) are comparable to all bladder or kidney cancers.