TAO.com

Claims 111 Verified account @DeSciClaims · 16h

22,706 UK adults who do not exercise wore wrist trackers.

Those whose trackers recorded four minutes of vigorous movement a day were diagnosed 36% less often with type 2 diabetes over 8 years.

But the study does not establish causality.

This paper went through Claims last week. The record includes 17 adjudicated claims, each tied to its evidence.

Each person wore the tracker for a week. Software read the raw signal and picked out bursts of vigorous movement lasting a minute or less, the kind that come from climbing stairs or walking briskly to the shop.

The diagnoses came from digital health records:
665 of the 22,706 developed type 2 diabetes.

Nobody was asked to move more. The trackers recorded what people already did. But people who move briskly through their day may differ from those who do not in ways no model captures.

The researchers, Emmanuel Stamatakis' group at the University of Sydney's Charles Perkins Centre @CPC_usyd, name a second caveat themselves: prediabetes that nobody had diagnosed when the study began.

They tested both. They left out anyone diagnosed in the first 2 years, then re-ran the numbers without people in poor health. The link held.

They still write that an observational study "cannot provide definitive evidence of the causality of the observed associations".

#MedTwitter

Chong et al. (2026), Diabetes Care.

doi.org/10.2337/dc25-3…

View quoted post on X

Beam - Subnet 105 Verified account @b1m_ai · 2d

Everyone is racing to build smarter models.

Meanwhile, the dataset is still sitting in the wrong cloud.
Beam is the open layer that moves data to where AI can actually use it.

Models get the glory.
Infrastructure gets the job done.

View quoted post on X

True Performance Network Verified account @TPN_Labs · 16h

TPN's 8th model is under production!

We shipped 7 optimized models on @huggingface, coming from different labs like @googlegemma, @Alibaba_Qwen and @MistralAI.

Competition n8 is the largest one that TPN has ever hosted, with the scope to take @Google's Gemma-4-31B under a 25GiB ceiling.

The rewards for the winner have increased, as the task at hand became more complex.

We wish good luck to the participants, and may the best model surface as winner.

View quoted post on X

Mark Jeffrey Verified account @markjeffrey · 1d

Bittensor $TAO just released quantum-resistant wallets.

View quoted post on X