1 in 5 comatose patients is aware

and why AI is not an existential threat to humanity

In partnership with

Welcome to this edition of Over a Cup of Coffee!

Before we head to our stories of this week, a BIG Thank you to the following subscribers for reading this newsletter without fail every week.

  • JP B

  • Ursa M

  • Sunil (at BS)

  • Shubha D

  • Siddharth B (at NH)

Your support means the world to me and keeps me going.

A friendly reminder again that this newsletter can now also be heard. Just click on the Listen Online Button on top

In this newsletter

Health apps are useful and deliver real results

A meta-analysis involving over 200,000 people from 47 studies has found that digital health tools like apps, text messages, and websites are actually beneficial in attaining the goals set out, whether it is achieving the number of steps or improving eating and sleeping behavior. 

With the economic burden of chronic health diseases expected to surpass US$47 trillion, effective interventions are the need of the hour and the study conducted by researchers at the University of South Australia now shows that digital tools are helpful. 

More specifically, these tools were effective in helping people achieve nearly an hour of moderate exercise every week, over 1,300 additional steps every day, and up to 20 percent more fruit and vegetable consumption in their diets. 

Interestingly, this behavior was found to be consistent across age groups, health populations and health behaviors. So, say yes to digital health apps and see your health improve. 

The research findings were published in the Nature’s Digital Medicine journal. 

AI can’t learn, Not an existential threat! 

Researchers at the University of Bath in the UK have conducted multiple analyses to showcase that artificial intelligence models like ChatGPT cannot learn on their own or acquire new skills. In short, they are not a threat to humanity in their current form. 

In certain cases, tools like ChatGPT show an ability to respond to social situations even though they have not been exlicitly trained on them. This has raised concerns if AI is getting smarter and could potentially outsmart humans one day. 

However, Harish Tayyar Madabushi, a computer scientist at Bath attributes this skill to ‘in-context learning’ which large language models (LLMs) can gain from following instructions, and their memory and linguistic proficiency. 

Madabushi is confident that even as LLMs are trained on larger sets of data, they will only be able to master one skill of language and not develop the ability to reason. 

Instead, what AI models are learning is clearly by instruction, which can be controlled and does not pose an existential risk to humans anytime soon. 

Research findings were published in ACL Anthology

All your news. None of the bias.

Be the smartest person in the room by reading 1440! Dive into 1440, where 3.5 million readers find their daily, fact-based news fix. We navigate through 100+ sources to deliver a comprehensive roundup from every corner of the internet – politics, global events, business, and culture, all in a quick, 5-minute newsletter. It's completely free and devoid of bias or political influence, ensuring you get the facts straight.

One in five comatose patients is aware

A study from the Weill Cornell Medical College in New York has now confirmed that one in five comatose patients are actually aware of their surroundings. This was confirmed by carrying out brain scans of 353 patients with severe brain damage, being maintained at six international centers of medicine. 

Previous research had found that about 10 percent of patients were in a state of “lock-in”, where their brain responds to instructions but are unable to communicate. 

To find out more, Nicholas Schiff and colleagues at Cornell Medical College, asked participants to think about performing physical activities like swimming or playing tennis for 15-30 seconds before pausing and were then asked think about the activity again. This was repeated seven times over five minutes, and MRI or electroencephalogram (EEG) data was recorded. 

Although 241 patients, showed no responses, brain activity for others matched that seen in volunteers with no brain damage. With nearly 400,000 patients with such consciousness disorders estimated worldwide, nearly 100,000 could be aware. 

This provides more information to families who need to make decisions about continuing life support treatments. With technologies like Brain-Computer Interfaces becoming more mainstream, the day will not be far when a communication technique can be attained with patients. 

The research findings were published in the New England Journal of Medicine

When do adults age? At 44 and then at 60! 

One might be celebrating their increased age every year but deep down inside the body, the ageing process happens much like growth - in bursts. Now, researchers at the Stanford University of Medicine have located these ageing bursts in mid-40s and then at 60. 

To arrive at this conclusion, the researchers collected samples of blood, saliva, stool, mouth, skin, and nose every few months from over 100 volunteers for up to seven years. By researching metabolites, RNA molecules, proteins, and microbes in these samples, the researchers found that these molecules saw sudden shifts at 44 and 60 instead of gradual shifts year on year. 

At first, it was thought that menopause in women was skewing the data for the shift during the 40s. But a closer look revealed that even male samples showed these shifts at the same age. 

In the mid-40s, the shift affected molecules related to cardiovascular disease and the ability to metabolize caffeine, lipids, and alcohol. In the 60s, the shift impacted molecules responsible for kidney function, immune regulation, and carbohydrate metabolism. 

Skin and muscle-related molecules saw a switch at both these age points. 

The research findings were published in the journal Nature Aging

Thanks for reading.
Until next time,
Ameya

Reply

or to participate.